System

Abstract

To provide a system capable of reducing the time and labor required for preparing the sentence of mail or chat and quickly preparing a consistent and high quality message.SOLUTION: The specification processing unit 290 of the data processing system 12 in the system executes a process of collecting past communication data of a user, a process of extracting a feature amount from the collected communication data, a process of training a AI model on the basis of the extracted feature amount, a process of generating a draft of a sentence in real time by using the trained AI model, and a process of displaying the generated draft to the user.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The technology of the present disclosure relates to a system. [Background technology]

[0002] Patent document 1 discloses a persona chatbot control method performed by at least one processor, the method including the steps of receiving a user utterance, adding the user utterance to a prompt including an instruction sentence related to a description of the chatbot character, encoding the prompt, and inputting the encoded prompt into a language model to generate a chatbot utterance in response to the user utterance. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-180282 Summary of the Invention [Problem to be solved by the invention]

[0004] In today's business and personal communications environments, it is common for email and chat writing to require a significant amount of time and effort. This can lead to variations in the quality of communication, making it difficult to quickly create consistent, high-quality messages. Furthermore, with the increase in remote work and non-face-to-face communication, efficient and natural writing is required. Furthermore, it is difficult to automatically generate drafts that reflect a user's own writing style, and it is not easy to imitate the writing style of others. [Means for solving the problem]

[0005] As a means for solving the above problems, the present invention provides a system having the following configuration.

[0006] The system includes a means for collecting a user's past communication data and extracting features from the collected communication data, a means for training an AI model based on the extracted features, a means for generating a draft of a sentence in real time using the trained AI model, and a means for displaying the generated draft to the user.

[0007] Furthermore, by adding a means for providing an interface that allows the user to correct and send the generated draft, the user can check the generated draft and make any necessary corrections before sending the message. Also, by adding a means for learning communication data of other people and imitating their writing style, the user can create sentences that reflect the style of others. This system makes sentence creation more efficient and achieves consistent, high-quality communication.

[0008] "User" refers to an individual or company that uses the system to create email or chat messages.

[0009] "Communication data" refers to data including a user's past email and chat message history.

[0010] "Features" refer to attributes such as word usage, punctuation patterns, and line break habits extracted from collected communication data.

[0011] "AI model" refers to an artificial intelligence algorithm or machine learning model that learns features and generates sentences that suit the user's writing style.

[0012] A "draft" refers to an initial version of an email or chat message generated by an AI model that reflects the user's writing style and habits.

[0013] "Interface" refers to the screens and tools that allow users to operate the system and review, correct, and submit the generated drafts.

[0014] "Send" refers to the act of a user sending a completed email or chat message to another person. [Brief explanation of the drawings]

[0015] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 3 is a sequence diagram showing a processing flow of the data processing system according to the first embodiment. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1. [Figure 13]FIG. 10 is a sequence diagram showing the flow of processing in the data processing system according to the second embodiment when an emotion engine is combined. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0016] An example of an embodiment of a system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

[0017] First, the terms used in the following description will be explained.

[0018] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be a single type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), and an APU (Accelerated Processing Unit).

[0019] In the following embodiments, a coded RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0020] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0021] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0022] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed connected by "and / or."

[0023] [First embodiment]

[0024] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0025] 1, a data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0026] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0027] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0028] The reception device 38 includes a touch panel 38A, a microphone 38B, and the like, and receives user input. The touch panel 38A detects contact with an indicator (for example, a pen or a finger) to receive user input by the touch of the indicator. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0029] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (for example, audio and / or text). The display 40A displays visible information such as text and images in accordance with instructions from the processor 46. The speaker 40B outputs audio in accordance with instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0030] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54.

[0031] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0032] 2, in the data processing device 12, a specific process is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0033] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0034] In the smart device 14, the processor 46 performs the reception output process. The storage 50 stores a reception output program 60. The reception output program 60 is used in conjunction with the specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0035] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0036] This invention is a system that collects users' past communication data, extracts features, and generates documents using AI technology. This system is designed to improve the efficiency of email and chat document creation and generate consistent, high-quality documents.

[0037] 1. Data Collection Module

[0038] User data integration

[0039] Users log in to the system and link their email and chat accounts. After linking, the user's device collects past email and chat history from these accounts and sends it to the server. Data transmission is encrypted to protect privacy.

[0040] 2. Model Learning Module

[0041] Data Preprocessing

[0042] The server preprocesses the communication data it receives. Specifically, it cleans and tokenizes the text to extract features such as the user's vocabulary, punctuation, and line break habits.

[0043] Training an AI model

[0044] The features are used to train an AI model. The server uses a machine learning algorithm (e.g., a Transformer model) to build a model that generates sentences that match the user's writing style.

[0045] 3. Sentence Generation Module

[0046] Draft Generation

[0047] When a user accesses the new message creation screen, the device sends a request to the server to generate a draft. The server uses a trained AI model to generate a draft in real time and sends it to the device. The generated draft reflects the user's writing style and habits.

[0048] 4. User Interface Module

[0049] View and edit drafts

[0050] The terminal displays the draft received from the server on the user interface. The user can check the displayed draft and make corrections or additions as necessary. The corrected message is completed on the terminal without being sent to the server again.

[0051] Sending a message

[0052] After the user has completed the edits, the final message is sent by clicking the "Send" button. The terminal then performs the process to send the final message.

[0053] Specific examples

[0054] For example, consider a situation where a business user wants to send an email proposing a new project. The user uses the system and uploads hundreds of past proposal emails to the system. The system learns from these emails and understands the user's writing style and proposal format. When the user creates a new proposal email, the system generates a draft like the one below.

[0055] Dear [Recipient Name],

[0056] Thank you for your continued support.

[0057] This is your [username] at [your company name].

[0058] Below we provide additional information regarding the new project we proposed the other day.

[0059] ...

[0060] If you have any questions, please feel free to contact us.

[0061] Sincerely,

[0062] [User Name]

[0063] Users can review this draft, add specific project details and dates, and then send it, allowing them to efficiently create consistent, high-quality messages.

[0064] The above is a specific embodiment for carrying out the present invention. This system significantly reduces the effort required for writing text and enables natural and effective communication.

[0065] The processing flow will be explained below.

[0066] Step 1:

[0067] A user logs in to the system and links their email account or chat account to the system. The user then enters authentication information to allow the linking.

[0068] Step 2:

[0069] The device retrieves past email and chat history from linked accounts. The device collects this data through an API and encrypts it for privacy reasons.

[0070] Step 3:

[0071] The device sends the encrypted data to the server, which receives it and stores it securely in a database. The data remains encrypted while it is stored.

[0072] Step 4:

[0073] The server retrieves the stored emails and chat history from the database and performs pre-processing, which includes cleaning and tokenization, to ensure the data is in a suitable format for analysis.

[0074] Step 5:

[0075] The server extracts features from the preprocessed data, including the user's vocabulary, punctuation patterns, and line-breaking habits.

[0076] Step 6:

[0077] The server uses the extracted features to train an AI model, using a machine learning algorithm (e.g., a Transformer model) to generate a model that matches the user's writing style. This process also includes model evaluation and optimization.

[0078] Step 7:

[0079] The user opens a new message composition screen. The device sends a request to the server to create a draft, including context information (subject and keywords).

[0080] Step 8:

[0081] The server uses a trained AI model to generate drafts in real time, which reflect the user's writing style and habits.

[0082] Step 9:

[0083] The server sends the generated draft to the terminal, which displays the draft on the user interface.

[0084] Step 10:

[0085] The user can check the displayed draft and make any necessary corrections or additions. The user can make these edits in real time using the device interface.

[0086] Step 11:

[0087] The user clicks the send button on the completed message, and the terminal constructs the final message and processes it for sending.

[0088] Step 12:

[0089] The terminal sends the final message to the server, which then sends the received message to the destination via a mail server or chat server.

[0090] Example 1

[0091] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0092] In today's business environment, users are required to maintain fast, consistent, and high-quality communications. However, manually creating documents is time-consuming and labor-intensive, making it difficult, especially when large volumes of communication are required. Privacy protection and data security are also important issues. Therefore, there is a need for a system that can effectively utilize users' past communication data and generate high-quality documents while safely managing the data.

[0093] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0094] In this invention, the server includes means for collecting past communication data of users, means for extracting features from the collected communication data, means for training an AI model based on the extracted features, means for generating drafts of sentences in real time using the trained AI model, means for displaying the generated drafts to the user, and means for protecting data using encryption technology, thereby enabling users to generate high-quality sentences quickly and consistently, while also ensuring privacy protection and data security.

[0095] "User's past communication data" is a collection of digital data related to communications such as emails and chat messages that the user has sent and received in the past.

[0096] "Features" refer to specific attributes and patterns extracted from collected communication data, including, for example, a user's vocabulary, punctuation, and line break habits.

[0097] An "AI model" is an artificial intelligence framework that has been trained using machine learning algorithms to perform a specific task (in this case, sentence generation).

[0098] "Training" refers to the process of having an AI model learn from collected data and acquire the ability to accomplish a given task.

[0099] A "draft" is an initial version of a message that a user is about to create, and is a basic sentence generated by an AI model.

[0100] "Encryption technology" refers to a technical method used to prevent data from being deciphered by third parties, and is used particularly to increase the security of communications.

[0101] A "user interface" is a display screen or operation panel that allows a user to interact with a system.

[0102] This invention is a system that collects users' past communication data, extracts features, and generates documents using AI technology. This system is designed to improve the efficiency of email and chat document creation and generate consistent, high-quality documents.

[0103] Data Collection Module

[0104] The user logs in to the system and links their email and chat accounts. The device collects past email and chat history from the linked accounts, encrypts it, and sends it to the server. AES encryption technology is used for encryption. The server securely stores the received data.

[0105] Model Learning Module

[0106] The server preprocesses the received communication data by cleaning and tokenizing it, and extracting features such as the user's vocabulary, punctuation, and line-breaking habits. To do this, it uses an NLP (Natural Language Processing) library (e.g., SpaCy or NLTK).

[0107] The server then trains an AI model based on the features. It uses a transformer model (e.g., BERT or GPT) and builds the model on a GPU cloud (e.g., NVIDIA GPU cloud). After the training process, the server creates a sentence generation model that matches the user's writing style.

[0108] Sentence Generation Module

[0109] When a user accesses the new message creation screen, the device sends a request to the server to generate a draft. The server uses a trained AI model to generate a draft in real time and sends it to the device. The generated draft reflects the user's writing style and habits.

[0110] User Interface Module

[0111] The terminal displays the draft received from the server on the user interface. The user can check the displayed draft and make corrections or additions as necessary. In this case, the corrected message is completed on the terminal without being resent to the server.

[0112] Finally, when the user clicks the "send" button, the device processes the final message to send it, allowing users to efficiently create consistent, high-quality messages.

[0113] Specific examples

[0114] For example, consider a situation where a business user wants to send an email proposing a new project. The user uses the system and uploads hundreds of past proposal emails to the system. The system learns from these emails and understands the user's writing style and proposal format. When the user creates a new proposal email, the system generates a draft like the one below.

[0115] Dear [Recipient Name],

[0116] Thank you for your continued support.

[0117] This is your [username] at [your company name].

[0118] Below we provide additional information regarding the new project we proposed the other day.

[0119] ...

[0120] If you have any questions, please feel free to contact us.

[0121] Sincerely,

[0122] [User Name]

[0123] Users can review this draft, add specific project details and dates, and then send it, allowing them to efficiently create consistent, high-quality messages.

[0124] The above is a specific embodiment for carrying out the present invention. This system significantly reduces the effort required for writing text and enables natural and effective communication.

[0125] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0126] Step 1:

[0127] User login and account linking

[0128] 1. A user logs into the system.

[0129] Input: User ID and password

[0130] Output: Authentication result (success / failure)

[0131] How it works: The user enters their ID and password on the dedicated login screen and clicks the login button. The server performs authentication, and if login is successful, their personal page is displayed.

[0132] 2. Users link their email and chat accounts to the system.

[0133] Input: Account information (email address, API key, etc.)

[0134] Output: Notification of integration success / failure

[0135] How it works: The user selects "Account Linking" from the settings screen and enters their account information. When they click the link button, the device will link their account and display a notification when the linking is complete.

[0136] Step 2:

[0137] Data Collection and Encryption

[0138] 1. Collect past email and chat history from the account linked to the device.

[0139] Input: Linked account information

[0140] Output: Collected communication data

[0141] How it works: The device uses APIs and other protocols to collect past email and chat history from your account. The collected data is temporarily stored on the device.

[0142] 2. The collected data is encrypted and sent to the server.

[0143] Input: communication data

[0144] Output: Encrypted data

[0145] How it works: Your device encrypts data using AES encryption and sends it to the server, which then decrypts it and stores it securely.

[0146] Step 3:

[0147] Data Preprocessing

[0148] 1. The server preprocesses the data it receives.

[0149] Input: Received encrypted data

[0150] Output: Preprocessed text data

[0151] How it works: The server decrypts the encrypted data and cleans it, specifically removing unnecessary metadata and duplicate messages.

[0152] 2. The server tokenizes the text and extracts features.

[0153] Input: Cleaned text data

[0154] Output: Tokenized data and features

[0155] How it works: The server uses an NLP (natural language processing) library (e.g., SpaCy or NLTK) to tokenize the text and extract features such as the user's vocabulary, punctuation, and line-breaking habits.

[0156] Step 4:

[0157] Training an AI model

[0158] 1. The server trains an AI model based on the features.

[0159] Input: Feature data

[0160] Output: A trained AI model

[0161] How it works: The server uses a Transformer model (e.g., BERT or GPT) to train an AI model on a GPU cloud (e.g., NVIDIA GPU cloud).

[0162] Step 5:

[0163] Real-time draft generation

[0164] 1. The user accesses the new message composition screen.

[0165] Input: User action (clicking New)

[0166] Output: Draft creation request

[0167] Behavior: The user clicks the "Create New" button on the message composition screen. The device sends a JSON-formatted draft creation request to the server.

[0168] 2. The server generates drafts in real time using a trained AI model.

[0169] Input: Draft generation request

[0170] Output: Generated draft

[0171] How it works: The server processes the draft generation request, generates the draft using the trained AI model, and returns the generated text in JSON format to the device.

[0172] Step 6:

[0173] Check and correct via user interface

[0174] 1. The terminal displays the received draft on the user interface.

[0175] Input: Generated draft data

[0176] Output: Draft text displayed

[0177] Operation: The device parses the received JSON data and displays it in the user interface via a text editor.

[0178] 2. The user reviews the draft and makes corrections or additions as necessary.

[0179] Input: Draft text displayed

[0180] Output: Corrected text

[0181] How it works: A user edits a draft in a text editor and then reviews the completed message.

[0182] Step 7:

[0183] Sending the final message

[0184] 1. The user clicks the "Submit" button.

[0185] Input: User action (click submit)

[0186] Output: Start of transmission process

[0187] Behavior: When the user clicks the "Send" button, the device processes the final message to send it.

[0188] 2. The device sends a message.

[0189] Input: Final modified message

[0190] Output: Message sent

[0191] What it does: Your device sends a message to the other person via email or chat account.

[0192] The above are the specific processing steps of the system program, which enables users to create and send messages efficiently, consistently, and with high quality.

[0193] (Application example 1)

[0194] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0195] Customer support in food delivery services requires quick and consistent responses, which places a heavy burden on customer support agents. Furthermore, when different agents respond, inconsistent writing styles and response policies can lead to lower customer satisfaction. To solve this problem, an efficient and consistent message generation system is needed.

[0196] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0197] In this invention, the server includes means for collecting past communication data of a user, means for extracting features from the collected communication data, means for training an AI model based on the extracted features, means for generating message drafts in real time using the trained AI model, means for displaying the generated drafts to the user, and means for automating customer support in a food delivery service, thereby enabling customer support personnel to efficiently generate consistent, high-quality messages and improving customer satisfaction.

[0198] "User" refers to the customer support staff of the food delivery service.

[0199] "Communication data" refers to records of chat and email exchanges between users and customers that have been collected in the past.

[0200] "Features" refer to the user's vocabulary and structural features extracted from communication data.

[0201] "AI model" refers to an artificial intelligence model that is trained based on features and learns the user's writing style and response patterns.

[0202] "Message Draft" refers to a draft response to a customer that is generated in real time by a trained AI model.

[0203] The "means for displaying" refers to a function for displaying the generated draft of the message on the user's terminal.

[0204] "Food delivery service" refers to all services that involve ordering and delivering meals.

[0205] "Customer response automation" refers to the ability for users to use AI models to automatically provide appropriate responses to customer inquiries and complaints.

[0206] The present invention provides a system for improving the efficiency of customer service in food delivery services and generating consistent, high-quality messages. Specific embodiments of this system are described below.

[0207] 1. Data Collection Module

[0208] The server collects users' past communication data. This communication data is a record of chats and emails exchanged between customer support representatives and customers. The data is encrypted and appropriate measures are taken to protect privacy. The specific software used is an API that uses the requests library to retrieve data from the server.

[0209] 2. Model Learning Module

[0210] The server preprocesses the collected communication data and extracts features, such as stylistic and structural characteristics, from the text. This involves text processing such as tokenization and cleaning. An AI model is then trained using specific machine learning algorithms, such as AutoTokenizer and AutoModelForSeq2SeqLM from the transformers library. This training builds a model that matches the user's writing style.

[0211] 3. Sentence Generation Module

[0212] The server uses a trained AI model to generate message drafts in real time. When a user enters a prompt, the AI ​​model generates an appropriate reply message draft and sends the result to the user's device. This allows for fast, high-quality messages.

[0213] 4. User Interface Module

[0214] The terminal displays the generated draft message to the user. The user can check the displayed draft and make corrections as necessary. The corrected message can then be sent to the customer as is. This allows the user to respond consistently without any hassle.

[0215] Specific examples

[0216] For example, consider the following prompt text entered by a customer support representative:

[0217] (prompt statement)

[0218] "Complaint handling: If a customer reports that there was a foreign object in the food they ordered, how would you respond?"

[0219] Based on this prompt, the AI ​​model generates a draft message like this:

[0220] (Generated Draft)

[0221] "Dear customers,

[0222] Thank you for using our food delivery service.

[0223] We deeply apologize for the report that a foreign object was found in the food you received.

[0224] We will investigate the matter immediately and respond promptly, so please wait a moment.

[0225] We will also give you a special discount on your next order.

[0226] Thank you for your continued support.

[0227] Sincerely,

[0228] Customer Support Representative Name

[0229] The system of the present invention enables customer support personnel to provide efficient and consistent responses, thereby improving customer satisfaction.

[0230] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0231] Step 1: User logs into the system

[0232] The user logs in to the system and links to their food delivery service account. At this time, they enter authentication information such as their ID and password. The system uses the entered authentication information to authenticate the user and allow them to log in. This makes it possible to access the user's past communication data.

[0233] Step 2: Collecting communication data

[0234] The server collects the user's past communication data through the API. Specifically, it uses the requests library to obtain the communication data using the API token and user ID. This data includes chat history with customers and email content. It receives the API token and user ID as input and obtains the communication data in JSON format as output.

[0235] Step 3: Preprocessing the data

[0236] The server preprocesses the collected communication data. This preprocessing includes cleaning and tokenizing the communication data, and uses AutoTokenizer from the transformers library. It receives JSON-formatted communication data as input and obtains clean text data as output. This makes it easier to extract noise-free features.

[0237] Step 4: Feature extraction

[0238] The server extracts features from the preprocessed communication data. Features refer to the user's writing style and structural characteristics, such as word frequency and sentence structure. It receives clean text data as input and obtains a feature vector as output. This feature vector is used for subsequent AI model training.

[0239] Step 5: Training the AI ​​model

[0240] The server trains an AI model based on the features. It uses AutoModelForSeq2SeqLM from the transformers library, receives the feature vector as input, and obtains a trained AI model as output. The trained AI model generates sentences that match the user's writing style.

[0241] Step 6: Enter a prompt statement

[0242] When a user creates a new message, they input a prompt text. For example, they can input a specific prompt text such as "Complaint handling: If a customer reports that a foreign object was found in the food they ordered, how would you respond?" The prompt text is received as input and sent to the system.

[0243] Step 7: Draft your message

[0244] The server generates a message draft in real time based on the input prompt. It uses the transformers library to process the prompt with an AI model. It takes the prompt as input and generates a message draft as output. This draft is adapted to the user's writing style.

[0245] Step 8: View and edit your draft

[0246] The terminal displays the generated message draft to the user. The user can review the displayed draft and make corrections as necessary. The corrected message is saved immediately and is not sent to the server again. It receives the message draft as input and displays it to the user as output.

[0247] Step 9: Sending a message

[0248] After the user completes the modification, the final message is sent. The terminal processes the final message and sends a quick reply to the customer. It receives the modified message as input and sends the final message to the customer as output.

[0249] This process allows customer support representatives at food delivery services to efficiently generate consistent, high-quality messages, improving customer satisfaction.

[0250] Furthermore, an emotion engine that estimates the user's emotion may be combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion.

[0251] This invention is a system that collects a user's past communication data, extracts features, and generates documents using AI technology. Furthermore, this invention incorporates an emotion engine that recognizes the user's emotions and adjusts the style and wording of the generated text based on those emotions. This system improves the efficiency of email and chat text creation and generates consistent, high-quality text.

[0252] 1. Data Collection Module

[0253] User data integration

[0254] Users log in to the system and link their email and chat accounts. After linking, the user's device collects past email and chat history from these accounts and sends it to the server. Data transmission is encrypted to protect privacy.

[0255] 2. Model Learning Module

[0256] Data Preprocessing

[0257] The server pre-processes the received communication data, which includes cleaning and tokenization, to make the data suitable for analysis.

[0258] Feature extraction

[0259] The server extracts features from the preprocessed data, including the user's vocabulary, punctuation patterns, and line-breaking habits.

[0260] Training an AI model

[0261] The server uses the extracted features to train an AI model, using a machine learning algorithm (e.g., a Transformer model) to generate a model that matches the user's writing style. This process also includes model evaluation and optimization.

[0262] 3. Sentence Generation Module

[0263] Draft Generation

[0264] When a user accesses the new message creation screen, the device sends a request to the server to generate a draft. The server uses a trained AI model to generate a draft in real time and sends it to the device. The generated draft reflects the user's writing style and habits.

[0265] 4. Emotion Engine

[0266] Emotion recognition

[0267] The device's built-in emotion engine recognizes the user's emotions through, for example, the user's typing speed, keyboard usage, and facial expression recognition.

[0268] Emotion-based writing style adjustment

[0269] The emotion engine adjusts the style and wording of the sentences generated by the AI ​​model based on the user's emotional state as recognized by the emotion engine. For example, if the user is excited, the language and style will be changed to a calmer one, and if the user is anxious, the language and style will be changed to a more reassuring one.

[0270] Real-time feedback

[0271] The emotion engine provides real-time feedback to users that reflects their emotions, allowing them to check whether their emotional state is reflected in their writing and make appropriate corrections.

[0272] 5. User Interface Module

[0273] View and edit drafts

[0274] The terminal displays the draft received from the server on the user interface. The user can check the displayed draft and make any necessary corrections or additions. The user can edit in real time.

[0275] Sending a message

[0276] After the user has completed the modifications, the final message is sent by clicking the "Send" button. The terminal will then construct the final message and process it for sending.

[0277] Specific examples

[0278] For example, consider a case where a user wants to send an important business email under emotional circumstances. The user logs in to the system, collects past business emails, and sends them to the server. The server learns from these and understands the user's writing style and suggested formats. When the user composes a new business email, the system recognizes the user's emotional state and generates a draft like the one below.

[0279] Dear [Recipient Name],

[0280] Thank you for your continued support.

[0281] This is your [username] at [your company name].

[0282] Below we provide additional information regarding the new project we proposed the other day.

[0283] ...

[0284] If you have any questions, please feel free to contact us.

[0285] Sincerely,

[0286] [User Name]

[0287] If the user is in a state of agitation, the system will adjust the tone of the sentence to a calmer and clearer tone, allowing the user to efficiently compose high-quality messages and ultimately achieve effective communication.

[0288] The above is a specific embodiment for carrying out the present invention. This system significantly reduces the effort required for writing text and enables natural and effective communication that takes into account the user's emotional state.

[0289] The processing flow will be explained below.

[0290] Step 1:

[0291] A user logs in to the system and links their email account or chat account to the system. The user then enters authentication information to allow the linking.

[0292] Step 2:

[0293] The device retrieves past email and chat history from linked accounts. The device collects this data through an API and encrypts it for privacy reasons.

[0294] Step 3:

[0295] The device sends the encrypted data to the server, which receives it and stores it securely in a database. The data remains encrypted while it is stored.

[0296] Step 4:

[0297] The server retrieves the stored emails and chat history from the database and performs pre-processing, which includes cleaning and tokenization, to ensure the data is in a suitable format for analysis.

[0298] Step 5:

[0299] The server extracts features from the preprocessed data, including the user's vocabulary, punctuation patterns, and line-breaking habits.

[0300] Step 6:

[0301] The server uses the extracted features to train an AI model, using a machine learning algorithm (e.g., a Transformer model) to generate a model that matches the user's writing style. This process also includes model evaluation and optimization.

[0302] Step 7:

[0303] The device's built-in emotion engine recognizes the user's emotions through their typing speed, keyboard usage, and facial expression recognition.

[0304] Step 8:

[0305] The user opens a new message composition screen. The device sends a request to the server to create a draft, including context information (subject and keywords).

[0306] Step 9:

[0307] The server uses a trained AI model to generate drafts in real time, which reflect the user's writing style and habits.

[0308] Step 10:

[0309] Based on the user's emotional state as recognized by the emotion engine, the server adjusts the writing style and language of the generated draft. For example, if the user is excited, the language and style will be changed to a calmer one.

[0310] Step 11:

[0311] The server generates and sends the adjusted draft to the terminal, which displays the draft on the user interface.

[0312] Step 12:

[0313] The user can check the displayed draft and make any necessary corrections or additions. The user can edit in real time.

[0314] Step 13:

[0315] The user clicks the send button on the completed message, and the terminal constructs the final message and processes it for sending.

[0316] Step 14:

[0317] The terminal sends the final message to the server, which then sends the received message to the destination via a mail server or chat server.

[0318] Example 2

[0319] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0320] Conventional writing support systems lack the ability to generate sentences that reflect the user's writing style and emotions, requiring the user to spend time manually correcting and adjusting them. Furthermore, sentences generated without considering the user's emotional state may result in a decline in the quality of communication. A system that solves these issues and generates high-quality sentences in real time that reflect the user's emotions while maintaining the user's writing style is needed.

[0321] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes means for collecting past communication data of a user, means for extracting features from the collected communication data, means for training an AI model based on the extracted features, means for generating a draft of a sentence in real time using the trained AI model, means for displaying the generated draft to the user, and means for recognizing the user's emotions and adjusting the style and wording of the generated sentence based on the user's emotional state. This enables real-time generation of high-quality sentences that reflect the user's style and emotions.

[0322] A "user" is an entity that uses this system to create, manage, and send communication data such as emails and chats.

[0323] "Communication data" refers to the content of emails and chats that a user has sent and received in the past.

[0324] "Features" are analytical units extracted from communication data, such as a user's unique language usage, punctuation patterns, and line break habits.

[0325] An "AI model" is an artificial intelligence algorithm trained based on features, which generates and predicts sentences.

[0326] The "emotion engine" is a function that recognizes the user's emotional state and adjusts the style and wording of the generated text according to that emotional state.

[0327] "Style adjustment" refers to changing the tone and expression of a sentence based on the user's emotional state and characteristics.

[0328] An "interface" is a screen or operating tool that allows a user to interact with the system, and provides functions for displaying, correcting, and sending drafts of text.

[0329] "Real-time" means responding immediately to user operations and immediately reflecting any results or changes that occur.

[0330] "Emotion recognition" refers to the technology or process of determining a user's emotional state based on their typing speed, keyboard usage, facial expressions, etc.

[0331] This invention is a system that collects a user's past communication data, extracts features, and generates documents using AI technology. It also includes an emotion engine that recognizes the user's emotions and adjusts the style and wording of the generated text accordingly. This system improves the efficiency of email and chat text creation and generates consistent, high-quality text.

[0332] User data integration

[0333] The user logs in to the system and links their email and chat accounts to the system. The user authenticates by entering their email address and password. Once the linking is complete, the device collects past email and chat history from these accounts via API and sends it to the server via encrypted communication.

[0334] Data Preprocessing

[0335] The server preprocesses the received communication data, cleansing it to remove noise and HTML tags, and then tokenizes it, breaking down sentences into words and converting them into a format suitable for analysis.

[0336] Feature extraction

[0337] The server extracts features from the preprocessed data, using natural language processing techniques to identify features such as the user's unique vocabulary, punctuation patterns, and line break habits.

[0338] Training an AI model

[0339] The server uses the extracted features to train an AI model, then uses a Transformer model as a machine learning algorithm to evaluate and optimize the model. This process generates the optimal model that best suits the user's writing style.

[0340] Draft Generation

[0341] When a user accesses the new message creation screen, the device sends a request to the server to generate a draft. The server uses a trained AI model to generate a draft in real time and sends it to the device. The generated draft reflects the user's writing style and habits.

[0342] Emotion recognition

[0343] The device's built-in emotion engine recognizes the user's emotions by analyzing their typing speed, keystroke patterns, and even facial expressions using a camera.

[0344] Emotion-based writing style adjustment

[0345] The emotion engine adjusts the style and language of the sentences generated by the AI ​​model based on the user's emotional state. For example, if the user is excited, the tone of the sentence will be changed to a calmer tone.

[0346] Real-time feedback

[0347] The device provides real-time feedback reflecting the user's emotions recognized by the emotion engine, allowing the user to see how their current emotional state is affecting their writing and make appropriate corrections.

[0348] View and edit drafts

[0349] The terminal displays the draft received from the server on the user interface. The user can check the displayed draft and make any necessary corrections or additions. The user can edit in real time and the changes are reflected immediately.

[0350] Sending a message

[0351] Once the user has finished editing the message, they click the send button to send the final message. The device then constructs the final message and sends it to the destination via the server in an encrypted connection.

[0352] Specific examples

[0353] For example, consider a case where a user wants to send an important business email under emotional circumstances. The user logs in to the system, collects past business emails, and sends them to the server. The server learns from these and understands the user's writing style and suggested formats. When the user composes a new business email, the system recognizes the user's emotional state and generates a draft like the one below.

[0354] Dear [Recipient Name],

[0355] Thank you for your continued support.

[0356] This is your [username] at [your company name].

[0357] Below we provide additional information regarding the new project we proposed the other day.

[0358] ...

[0359] If you have any questions, please feel free to contact us.

[0360] Sincerely,

[0361] [User Name]

[0362] If the user is in a state of agitation, the system can adjust the tone of the sentence to a calmer and clearer tone, allowing the user to efficiently compose high-quality messages and ultimately achieve effective communication.

[0363] The above is a specific embodiment for carrying out the present invention. This system significantly reduces the effort required for writing text and enables natural and effective communication that takes into account the user's emotional state.

[0364] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0365] Divide the program's processing flow into processing steps

[0366] Step 1: Linking user data

[0367] Step 2: Preprocessing the data

[0368] Step 3: Feature extraction

[0369] Step 4: Training the AI ​​model

[0370] Step 5: Draft generation

[0371] Step 6: Recognize emotions

[0372] Step 7: Adjust your writing style based on emotion

[0373] Step 8: Real-time feedback

[0374] Step 9: View and edit your draft

[0375] Step 10: Sending a message

[0376] Specific explanation of each step

[0377] Step 1: Linking user data

[0378] A user logs in to the system.

[0379] The user enters an email address and password to log in to the system and is authenticated.

[0380] Input: Email address, password

[0381] Output: User's authentication token

[0382] The device is linked to the user's email and chat accounts.

[0383] With the user's permission, the device accesses email and chat accounts through an API.

[0384] Input: Authentication token, account information

[0385] Output: Access rights

[0386] The collected data is sent to a server.

[0387] The device encrypts the collected past emails and chat history and sends it to the server.

[0388] Input: Past communication data

[0389] Output: Encrypted data sent to the server

[0390] Step 2: Preprocessing the data

[0391] The server preprocesses the data received.

[0392] The server cleans the data it receives, removing noise and HTML tags, and then tokenizes it to convert it into a format that is easier to analyze.

[0393] Input: Received communication data

[0394] Output: Preprocessed data

[0395] Step 3: Feature extraction

[0396] The server extracts features from the preprocessed data.

[0397] The server uses natural language processing technology to extract features such as the user's unique vocabulary, punctuation patterns, and line break habits.

[0398] Input: Preprocessed data

[0399] Output: Extracted features

[0400] Step 4: Training the AI ​​model

[0401] The server uses the extracted features to train an AI model.

[0402] The server uses the Transformer model as a machine learning algorithm to evaluate and optimize the model.

[0403] Input: extracted features

[0404] Output: A trained AI model

[0405] Step 5: Draft generation

[0406] The user accesses a new message composition screen.

[0407] The user logs into the system interface using a terminal and opens a message composition screen.

[0408] Input: User interface access

[0409] Output: Message creation screen

[0410] The terminal sends a request to the server to generate a draft.

[0411] The device sends a request to the server to generate a draft using the trained AI model.

[0412] Input: Draft generation request

[0413] Output: A draft creation request is sent to the server

[0414] The server sends the generated draft to the device.

[0415] The server returns the draft generated by the AI ​​model to the device.

[0416] Input: A trained AI model

[0417] Output: Draft document

[0418] Step 6: Recognize emotions

[0419] The emotion engine installed in the device recognizes the user's emotions.

[0420] The device analyzes keystroke speed, typing patterns and facial expressions using a camera to determine the user's emotions.

[0421] Input: key input data, keystroke patterns, facial expression data

[0422] Output: Recognized user emotion data

[0423] Step 7: Adjust your writing style based on emotion

[0424] The emotion engine adjusts writing style and language based on the emotional state it recognizes.

[0425] The server uses AI models to, for example, change the tone of the text to a calmer tone if the user is in an agitated state.

[0426] Input: Recognized emotion data, draft text

[0427] Output: Adjusted sentence

[0428] Step 8: Real-time feedback

[0429] The device provides real-time feedback reflecting the user's emotions recognized by the emotion engine.

[0430] The device displays emotional feedback on the screen, allowing users to see how their current emotional state is affecting their writing.

[0431] Input: Recognized emotion data

[0432] Output: Emotional feedback to the user

[0433] Step 9: View and edit your draft

[0434] The terminal displays the draft received from the server on the user interface.

[0435] The device provides an interface that allows you to edit the displayed draft as it is.

[0436] Input: Draft text

[0437] Output: Editable interface

[0438] The user checks the displayed draft and makes any necessary corrections or additions.

[0439] Users can edit in real time and see changes reflected immediately.

[0440] Input: Draft text

[0441] Output: Final text with corrections and additions

[0442] Step 10: Sending a message

[0443] The user clicks the send button to send the message.

[0444] After the user has completed the modifications, the final message is sent by clicking the send button.

[0445] Input: Final text with corrections and additions

[0446] Output: Send request

[0447] The terminal constructs the final message and sends it to the destination via the server.

[0448] The device encrypts the message according to the destination protocol and sends it to the intended recipient via the server.

[0449] Input: Send request, destination information

[0450] Output: Message sent

[0451] The above are the specific processing steps and their respective operations of this system, which make the writing process more efficient and realize natural communication that reflects the user's emotional state.

[0452] (Application example 2)

[0453] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0454] Customer support plays an important role in modern electronic payment services. However, while prompt and appropriate responses are required for user inquiries and troubleshooting, they often require a great deal of time and effort. Providing an efficient response is particularly difficult when it is necessary to take into account the user's emotional state. The purpose of this invention is to solve these problems and automatically generate prompt and emotionally sensitive responses.

[0455] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for collecting past communication data of a user, means for extracting features from the collected communication data, means for training an AI model based on the extracted features, means for generating drafts of sentences in real time using the trained AI model, means for displaying the generated drafts to the user, means for grasping the user's emotional state, and means for adjusting the style and wording of the generated sentences based on the grasped emotional state. This enables high-quality response sentences to be generated quickly and with consideration for emotions.

[0456] "User's past communication data" refers to the communication history of the user, such as emails and chats that the user has sent and received in the past.

[0457] "Features" are important attributes or patterns extracted from data to analyze the information.

[0458] An "AI model" is a computational model that uses artificial intelligence technology to learn from data and make predictions and classifications.

[0459] "Generating a draft of a text in real time" means creating a first draft of a text based on data that is instantly input.

[0460] "Displaying the generated draft to the user" means displaying the first draft generated by AI on the user's device screen.

[0461] Understanding the "user's emotional state" means observing and analyzing the user's current mental and emotional state.

[0462] "Adjusting style and language" means changing the way you write and express things appropriately.

[0463] An "interface" refers to the operation screen and input device that allow the user to interact with the system.

[0464] "Other people's communication data" refers to communication history such as emails and chats sent and received in the past by individuals other than the user.

[0465] "Imitating a style" means reproducing the writing style or expression of a particular person.

[0466] This invention is a system for automatically generating optimal responses to user inquiries and problems in customer support for electronic payment services. This system takes into consideration the user's emotional state and can generate quick, high-quality responses.

[0467] Overall system overview

[0468] The system primarily uses the following hardware and software components:

[0469] server

[0470] User's device (PC or smartphone)

[0471] Natural language processing libraries (e.g. spaCy or NLTK)

[0472] Transformer models (e.g., BERT and GPT-3)

[0473] Emotion recognition software (e.g., face-api.js)

[0474] Data Collection Module

[0475] When a user logs in to the service, the system collects the user's past communication data, including past emails and chat history. This data is securely transmitted to the server using SSL / TLS encryption protocols.

[0476] Model Learning Module

[0477] The server preprocesses the incoming communication data. This includes cleaning and tokenizing the data using an automated NLP library (e.g., spaCy or NLTK). Next, it extracts features from the data (e.g., wording and punctuation patterns). These features are used to retrain a Transformer model (e.g., BERT or GPT-3) to match the writing style of a specific user.

[0478] Emotion Recognition Module

[0479] The user's device monitors the user's typing speed and patterns in real time, and in some cases analyzes the user's emotional state using facial recognition technology (e.g., face-api.js) using a webcam. This information is used to understand the user's current emotional state (e.g., impatience, anxiety).

[0480] Text generation and style adjustment module

[0481] The trained AI model generates real-time draft responses to user queries, which are then sent from the server to the user's device, while adjusting the writing style and wording based on data obtained from the emotion recognition module.

[0482] User Interface Module

[0483] The generated draft is displayed on the user's screen, and the user can make corrections or additions in real time. The final confirmed message is sent by clicking the "Send" button.

[0484] Examples of specific examples and prompts

[0485] For example, if a user reports a problem with a "payment not being posted," the following response might be generated:

[0486] Dear (Username),

[0487] Thank you for your help. This is (Name of person in charge) from the customer support department of the electronic payment service.

[0488] We will look into the issue of your reported payment not being reflected as soon as possible. We apologize for any inconvenience this may cause.

[0489] Could you please provide the following information:

[0490] Transaction ID at the time of payment

[0491] Payment date and time

[0492] Payment amount

[0493] We will respond promptly based on the information you provide.

[0494] Thank you for your cooperation.

[0495] Sincerely,

[0496] (Name of person in charge)

[0497] If the user is in a state of agitation, the tone of the text will be adjusted to a calmer tone.

[0498] Example prompt sentence:

[0499] "Generate a response for a situation where a user is reporting difficulty with a payment and the user is in a panicked state. Consider the user's past inquiry history."

[0500] This system will significantly improve the efficiency and quality of customer support for electronic payment services.

[0501] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0502] Step 1:

[0503] A user logs in to the service. At this time, the user links their email account and chat account to the system. After linking, the user's device collects past email and chat history from these accounts. The collected data is encrypted and sent to the server. The input is the user's login information and communication data, and the output is the transmission of encrypted data.

[0504] Step 2:

[0505] The server preprocesses the communication data it receives. Specifically, it cleans and tokenizes the data using a natural language processing library (such as spaCy or NLTK). During this process, it removes unnecessary symbols and spaces and splits the data into words. The input is the decrypted communication data, and the output is the preprocessed, clean data.

[0506] Step 3:

[0507] The server extracts features from the preprocessed data. Features include the user's vocabulary, punctuation patterns, and line break habits. A feature extraction algorithm is used for extraction. The input is the preprocessed clean data, and the output is the features.

[0508] Step 4:

[0509] The server uses the extracted features to train an AI model. A Transformer model (e.g., BERT or GPT-3) is retrained with the specific user's data. This process also includes model evaluation and optimization. The input is the features, and the output is the trained AI model.

[0510] Step 5:

[0511] The user's device monitors the user's input speed and patterns in real time to grasp their emotional state. In some cases, a webcam is used to recognize facial expressions and analyze the user's emotions. The input is the user's input behavior and facial image, and the output is data on the user's emotional state.

[0512] Step 6:

[0513] When a user accesses the new message composition screen, the device sends a request to the server to generate a draft. The server uses a trained AI model to generate the draft in real time. The draft is based on the user's past communication data. The input is the user's request, and the output is the initial draft.

[0514] Step 7:

[0515] The server adjusts the writing style and language of the draft generated by the server based on data obtained from the emotion recognition module. For example, if the user is in a state of agitation, the writing style will be made more calm. The input is the initial draft and emotion data, and the output is the adjusted draft.

[0516] Step 8:

[0517] The generated draft is displayed on the user's device. The user can review the displayed draft and make corrections or additions as needed. This correction is done in real time. The input is the adjusted draft, and the output is the final message as modified by the user.

[0518] Step 9:

[0519] After the user completes the modifications, they click the "Send" button to send the final message. The terminal constructs the final message and processes it for sending. The input is the final message modified by the user, and the output is the sent message.

[0520] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0521] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0522] In the above embodiment, an example in which the specific process is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific process may be performed by the smart device 14.

[0523] [Second embodiment]

[0524] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0525] 3, the data processing system 210 includes the data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0526] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0527] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, and the camera 42 are also connected to the bus 52.

[0528] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0529] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0530] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0531] Fig. 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Fig. 4, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0532] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0533] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0534] In the smart glasses 214, the reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0535] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal."

[0536] This invention is a system that collects users' past communication data, extracts features, and generates documents using AI technology. This system is designed to improve the efficiency of email and chat document creation and generate consistent, high-quality documents.

[0537] 1. Data Collection Module

[0538] User data integration

[0539] Users log in to the system and link their email and chat accounts. After linking, the user's device collects past email and chat history from these accounts and sends it to the server. Data transmission is encrypted to protect privacy.

[0540] 2. Model Learning Module

[0541] Data Preprocessing

[0542] The server preprocesses the communication data it receives. Specifically, it cleans and tokenizes the text to extract features such as the user's vocabulary, punctuation, and line break habits.

[0543] Training an AI model

[0544] The features are used to train an AI model. The server uses a machine learning algorithm (e.g., a Transformer model) to build a model that generates sentences that match the user's writing style.

[0545] 3. Sentence Generation Module

[0546] Draft Generation

[0547] When a user accesses the new message creation screen, the device sends a request to the server to generate a draft. The server uses a trained AI model to generate a draft in real time and sends it to the device. The generated draft reflects the user's writing style and habits.

[0548] 4. User Interface Module

[0549] View and edit drafts

[0550] The terminal displays the draft received from the server on the user interface. The user can check the displayed draft and make corrections or additions as necessary. The corrected message is completed on the terminal without being sent to the server again.

[0551] Sending a message

[0552] After the user has completed the edits, the final message is sent by clicking the "Send" button. The terminal then performs the process to send the final message.

[0553] Specific examples

[0554] For example, consider a situation where a business user wants to send an email proposing a new project. The user uses the system and uploads hundreds of past proposal emails to the system. The system learns from these emails and understands the user's writing style and proposal format. When the user creates a new proposal email, the system generates a draft like the one below.

[0555] Dear [Recipient Name],

[0556] Thank you for your continued support.

[0557] This is your [username] at [your company name].

[0558] Below we provide additional information regarding the new project we proposed the other day.

[0559] ...

[0560] If you have any questions, please feel free to contact us.

[0561] Sincerely,

[0562] [User Name]

[0563] Users can review this draft, add specific project details and dates, and then send it, allowing them to efficiently create consistent, high-quality messages.

[0564] The above is a specific embodiment for carrying out the present invention. This system significantly reduces the effort required for writing text and enables natural and effective communication.

[0565] The processing flow will be explained below.

[0566] Step 1:

[0567] A user logs in to the system and links their email account or chat account to the system. The user then enters authentication information to allow the linking.

[0568] Step 2:

[0569] The device retrieves past email and chat history from linked accounts. The device collects this data through an API and encrypts it for privacy reasons.

[0570] Step 3:

[0571] The device sends the encrypted data to the server, which receives it and stores it securely in a database. The data remains encrypted while it is stored.

[0572] Step 4:

[0573] The server retrieves the stored emails and chat history from the database and performs pre-processing, which includes cleaning and tokenization, to ensure the data is in a suitable format for analysis.

[0574] Step 5:

[0575] The server extracts features from the preprocessed data, including the user's vocabulary, punctuation patterns, and line-breaking habits.

[0576] Step 6:

[0577] The server uses the extracted features to train an AI model, using a machine learning algorithm (e.g., a Transformer model) to generate a model that matches the user's writing style. This process also includes model evaluation and optimization.

[0578] Step 7:

[0579] The user opens a new message composition screen. The device sends a request to the server to create a draft, including context information (subject and keywords).

[0580] Step 8:

[0581] The server uses a trained AI model to generate drafts in real time, which reflect the user's writing style and habits.

[0582] Step 9:

[0583] The server sends the generated draft to the terminal, which displays the draft on the user interface.

[0584] Step 10:

[0585] The user can check the displayed draft and make any necessary corrections or additions. The user can make these edits in real time using the device interface.

[0586] Step 11:

[0587] The user clicks the send button on the completed message, and the terminal constructs the final message and processes it for sending.

[0588] Step 12:

[0589] The terminal sends the final message to the server, which then sends the received message to the destination via a mail server or chat server.

[0590] Example 1

[0591] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0592] In today's business environment, users are required to maintain fast, consistent, and high-quality communications. However, manually creating documents is time-consuming and labor-intensive, making it difficult, especially when large volumes of communication are required. Privacy protection and data security are also important issues. Therefore, there is a need for a system that can effectively utilize users' past communication data and generate high-quality documents while safely managing the data.

[0593] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0594] In this invention, the server includes means for collecting past communication data of users, means for extracting features from the collected communication data, means for training an AI model based on the extracted features, means for generating drafts of sentences in real time using the trained AI model, means for displaying the generated drafts to the user, and means for protecting data using encryption technology, thereby enabling users to generate high-quality sentences quickly and consistently, while also ensuring privacy protection and data security.

[0595] "User's past communication data" is a collection of digital data related to communications such as emails and chat messages that the user has sent and received in the past.

[0596] "Features" refer to specific attributes and patterns extracted from collected communication data, including, for example, a user's vocabulary, punctuation, and line break habits.

[0597] An "AI model" is an artificial intelligence framework that has been trained using machine learning algorithms to perform a specific task (in this case, sentence generation).

[0598] "Training" refers to the process of having an AI model learn from collected data and acquire the ability to accomplish a given task.

[0599] A "draft" is an initial version of a message that a user is about to create, and is a basic sentence generated by an AI model.

[0600] "Encryption technology" refers to a technical method used to prevent data from being deciphered by third parties, and is used particularly to increase the security of communications.

[0601] A "user interface" is a display screen or operation panel that allows a user to interact with a system.

[0602] This invention is a system that collects users' past communication data, extracts features, and generates documents using AI technology. This system is designed to improve the efficiency of email and chat document creation and generate consistent, high-quality documents.

[0603] Data Collection Module

[0604] The user logs in to the system and links their email and chat accounts. The device collects past email and chat history from the linked accounts, encrypts it, and sends it to the server. AES encryption technology is used for encryption. The server securely stores the received data.

[0605] Model Learning Module

[0606] The server preprocesses the received communication data by cleaning and tokenizing it, and extracting features such as the user's vocabulary, punctuation, and line-breaking habits. To do this, it uses an NLP (Natural Language Processing) library (e.g., SpaCy or NLTK).

[0607] The server then trains an AI model based on the features. It uses a transformer model (e.g., BERT or GPT) and builds the model on a GPU cloud (e.g., NVIDIA GPU cloud). After the training process, the server creates a sentence generation model that matches the user's writing style.

[0608] Sentence Generation Module

[0609] When a user accesses the new message creation screen, the device sends a request to the server to generate a draft. The server uses a trained AI model to generate a draft in real time and sends it to the device. The generated draft reflects the user's writing style and habits.

[0610] User Interface Module

[0611] The terminal displays the draft received from the server on the user interface. The user can check the displayed draft and make corrections or additions as necessary. In this case, the corrected message is completed on the terminal without being resent to the server.

[0612] Finally, when the user clicks the "send" button, the device processes the final message to send it, allowing users to efficiently create consistent, high-quality messages.

[0613] Specific examples

[0614] For example, consider a situation where a business user wants to send an email proposing a new project. The user uses the system and uploads hundreds of past proposal emails to the system. The system learns from these emails and understands the user's writing style and proposal format. When the user creates a new proposal email, the system generates a draft like the one below.

[0615] Dear [Recipient Name],

[0616] Thank you for your continued support.

[0617] This is your [username] at [your company name].

[0618] Below we provide additional information regarding the new project we proposed the other day.

[0619] ...

[0620] If you have any questions, please feel free to contact us.

[0621] Sincerely,

[0622] [User Name]

[0623] Users can review this draft, add specific project details and dates, and then send it, allowing them to efficiently create consistent, high-quality messages.

[0624] The above is a specific embodiment for carrying out the present invention. This system significantly reduces the effort required for writing text and enables natural and effective communication.

[0625] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0626] Step 1:

[0627] User login and account linking

[0628] 1. A user logs into the system.

[0629] Input: User ID and password

[0630] Output: Authentication result (success / failure)

[0631] How it works: The user enters their ID and password on the dedicated login screen and clicks the login button. The server performs authentication, and if login is successful, their personal page is displayed.

[0632] 2. Users link their email and chat accounts to the system.

[0633] Input: Account information (email address, API key, etc.)

[0634] Output: Notification of integration success / failure

[0635] How it works: The user selects "Account Linking" from the settings screen and enters their account information. When they click the link button, the device will link their account and display a notification when the linking is complete.

[0636] Step 2:

[0637] Data Collection and Encryption

[0638] 1. Collect past email and chat history from the account linked to the device.

[0639] Input: Linked account information

[0640] Output: Collected communication data

[0641] How it works: The device uses APIs and other protocols to collect past email and chat history from your account. The collected data is temporarily stored on the device.

[0642] 2. The collected data is encrypted and sent to the server.

[0643] Input: communication data

[0644] Output: Encrypted data

[0645] How it works: Your device encrypts data using AES encryption and sends it to the server, which then decrypts it and stores it securely.

[0646] Step 3:

[0647] Data Preprocessing

[0648] 1. The server preprocesses the data it receives.

[0649] Input: Received encrypted data

[0650] Output: Preprocessed text data

[0651] How it works: The server decrypts the encrypted data and cleans it, specifically removing unnecessary metadata and duplicate messages.

[0652] 2. The server tokenizes the text and extracts features.

[0653] Input: Cleaned text data

[0654] Output: Tokenized data and features

[0655] How it works: The server uses an NLP (natural language processing) library (e.g., SpaCy or NLTK) to tokenize the text and extract features such as the user's vocabulary, punctuation, and line-breaking habits.

[0656] Step 4:

[0657] Training an AI model

[0658] 1. The server trains an AI model based on the features.

[0659] Input: Feature data

[0660] Output: A trained AI model

[0661] How it works: The server uses a Transformer model (e.g., BERT or GPT) to train an AI model on a GPU cloud (e.g., NVIDIA GPU cloud).

[0662] Step 5:

[0663] Real-time draft generation

[0664] 1. The user accesses the new message composition screen.

[0665] Input: User action (clicking New)

[0666] Output: Draft creation request

[0667] Behavior: The user clicks the "Create New" button on the message composition screen. The device sends a JSON-formatted draft creation request to the server.

[0668] 2. The server generates drafts in real time using a trained AI model.

[0669] Input: Draft generation request

[0670] Output: Generated draft

[0671] How it works: The server processes the draft generation request, generates the draft using the trained AI model, and returns the generated text in JSON format to the device.

[0672] Step 6:

[0673] Check and correct via user interface

[0674] 1. The terminal displays the received draft on the user interface.

[0675] Input: Generated draft data

[0676] Output: Draft text displayed

[0677] Operation: The device parses the received JSON data and displays it in the user interface via a text editor.

[0678] 2. The user reviews the draft and makes corrections or additions as necessary.

[0679] Input: Draft text displayed

[0680] Output: Corrected text

[0681] How it works: A user edits a draft in a text editor and then reviews the completed message.

[0682] Step 7:

[0683] Sending the final message

[0684] 1. The user clicks the "Submit" button.

[0685] Input: User action (click submit)

[0686] Output: Start of transmission process

[0687] Behavior: When the user clicks the "Send" button, the device processes the final message to send it.

[0688] 2. The device sends a message.

[0689] Input: Final modified message

[0690] Output: Message sent

[0691] What it does: Your device sends a message to the other person via email or chat account.

[0692] The above are the specific processing steps of the system program, which enables users to create and send messages efficiently, consistently, and with high quality.

[0693] (Application example 1)

[0694] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0695] Customer support in food delivery services requires quick and consistent responses, which places a heavy burden on customer support agents. Furthermore, when different agents respond, inconsistent writing styles and response policies can lead to lower customer satisfaction. To solve this problem, an efficient and consistent message generation system is needed.

[0696] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0697] In this invention, the server includes means for collecting past communication data of a user, means for extracting features from the collected communication data, means for training an AI model based on the extracted features, means for generating message drafts in real time using the trained AI model, means for displaying the generated drafts to the user, and means for automating customer support in a food delivery service, thereby enabling customer support personnel to efficiently generate consistent, high-quality messages and improving customer satisfaction.

[0698] "User" refers to the customer support staff of the food delivery service.

[0699] "Communication data" refers to records of chat and email exchanges between users and customers that have been collected in the past.

[0700] "Features" refer to the user's vocabulary and structural features extracted from communication data.

[0701] "AI model" refers to an artificial intelligence model that is trained based on features and learns the user's writing style and response patterns.

[0702] "Message Draft" refers to a draft response to a customer that is generated in real time by a trained AI model.

[0703] The "means for displaying" refers to a function for displaying the generated draft of the message on the user's terminal.

[0704] "Food delivery service" refers to all services that involve ordering and delivering meals.

[0705] "Customer response automation" refers to the ability for users to use AI models to automatically provide appropriate responses to customer inquiries and complaints.

[0706] The present invention provides a system for improving the efficiency of customer service in food delivery services and generating consistent, high-quality messages. Specific embodiments of this system are described below.

[0707] 1. Data Collection Module

[0708] The server collects users' past communication data. This communication data is a record of chats and emails exchanged between customer support representatives and customers. The data is encrypted and appropriate measures are taken to protect privacy. The specific software used is an API that uses the requests library to retrieve data from the server.

[0709] 2. Model Learning Module

[0710] The server preprocesses the collected communication data and extracts features, such as stylistic and structural characteristics, from the text. This involves text processing such as tokenization and cleaning. An AI model is then trained using specific machine learning algorithms, such as AutoTokenizer and AutoModelForSeq2SeqLM from the transformers library. This training builds a model that matches the user's writing style.

[0711] 3. Sentence Generation Module

[0712] The server uses a trained AI model to generate message drafts in real time. When a user enters a prompt, the AI ​​model generates an appropriate reply message draft and sends the result to the user's device. This allows for fast, high-quality messages.

[0713] 4. User Interface Module

[0714] The terminal displays the generated draft message to the user. The user can check the displayed draft and make corrections as necessary. The corrected message can then be sent to the customer as is. This allows the user to respond consistently without any hassle.

[0715] Specific examples

[0716] For example, consider the following prompt text entered by a customer support representative:

[0717] (prompt statement)

[0718] "Complaint handling: If a customer reports that there was a foreign object in the food they ordered, how would you respond?"

[0719] Based on this prompt, the AI ​​model generates a draft message like this:

[0720] (Generated Draft)

[0721] "Dear customers,

[0722] Thank you for using our food delivery service.

[0723] We deeply apologize for the report that a foreign object was found in the food you received.

[0724] We will investigate the matter immediately and respond promptly, so please wait a moment.

[0725] We will also give you a special discount on your next order.

[0726] Thank you for your continued support.

[0727] Sincerely,

[0728] Customer Support Representative Name

[0729] The system of the present invention enables customer support personnel to provide efficient and consistent responses, thereby improving customer satisfaction.

[0730] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0731] Step 1: User logs into the system

[0732] The user logs in to the system and links to their food delivery service account. At this time, they enter authentication information such as their ID and password. The system uses the entered authentication information to authenticate the user and allow them to log in. This makes it possible to access the user's past communication data.

[0733] Step 2: Collecting communication data

[0734] The server collects the user's past communication data through the API. Specifically, it uses the requests library to obtain the communication data using the API token and user ID. This data includes chat history with customers and email content. It receives the API token and user ID as input and obtains the communication data in JSON format as output.

[0735] Step 3: Preprocessing the data

[0736] The server preprocesses the collected communication data. This preprocessing includes cleaning and tokenizing the communication data, and uses AutoTokenizer from the transformers library. It receives JSON-formatted communication data as input and obtains clean text data as output. This makes it easier to extract noise-free features.

[0737] Step 4: Feature extraction

[0738] The server extracts features from the preprocessed communication data. Features refer to the user's writing style and structural characteristics, such as word frequency and sentence structure. It receives clean text data as input and obtains a feature vector as output. This feature vector is used for subsequent AI model training.

[0739] Step 5: Training the AI ​​model

[0740] The server trains an AI model based on the features. It uses AutoModelForSeq2SeqLM from the transformers library, receives the feature vector as input, and obtains a trained AI model as output. The trained AI model generates sentences that match the user's writing style.

[0741] Step 6: Enter a prompt statement

[0742] When a user creates a new message, they input a prompt text. For example, they can input a specific prompt text such as "Complaint handling: If a customer reports that a foreign object was found in the food they ordered, how would you respond?" The prompt text is received as input and sent to the system.

[0743] Step 7: Draft your message

[0744] The server generates a message draft in real time based on the input prompt. It uses the transformers library to process the prompt with an AI model. It takes the prompt as input and generates a message draft as output. This draft is adapted to the user's writing style.

[0745] Step 8: View and edit your draft

[0746] The terminal displays the generated message draft to the user. The user can review the displayed draft and make corrections as necessary. The corrected message is saved immediately and is not sent to the server again. It receives the message draft as input and displays it to the user as output.

[0747] Step 9: Sending a message

[0748] After the user completes the modification, the final message is sent. The terminal processes the final message and sends a quick reply to the customer. It receives the modified message as input and sends the final message to the customer as output.

[0749] This process allows customer support representatives at food delivery services to efficiently generate consistent, high-quality messages, improving customer satisfaction.

[0750] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0751] This invention is a system that collects a user's past communication data, extracts features, and generates documents using AI technology. Furthermore, this invention incorporates an emotion engine that recognizes the user's emotions and adjusts the style and wording of the generated text based on those emotions. This system improves the efficiency of email and chat text creation and generates consistent, high-quality text.

[0752] 1. Data Collection Module

[0753] User data integration

[0754] Users log in to the system and link their email and chat accounts. After linking, the user's device collects past email and chat history from these accounts and sends it to the server. Data transmission is encrypted to protect privacy.

[0755] 2. Model Learning Module

[0756] Data Preprocessing

[0757] The server pre-processes the received communication data, which includes cleaning and tokenization, to make the data suitable for analysis.

[0758] Feature extraction

[0759] The server extracts features from the preprocessed data, including the user's vocabulary, punctuation patterns, and line-breaking habits.

[0760] Training an AI model

[0761] The server uses the extracted features to train an AI model, using a machine learning algorithm (e.g., a Transformer model) to generate a model that matches the user's writing style. This process also includes model evaluation and optimization.

[0762] 3. Sentence Generation Module

[0763] Draft Generation

[0764] When a user accesses the new message creation screen, the device sends a request to the server to generate a draft. The server uses a trained AI model to generate a draft in real time and sends it to the device. The generated draft reflects the user's writing style and habits.

[0765] 4. Emotion Engine

[0766] Emotion recognition

[0767] The device's built-in emotion engine recognizes the user's emotions through, for example, the user's typing speed, keyboard usage, and facial expression recognition.

[0768] Emotion-based writing style adjustment

[0769] The emotion engine adjusts the style and wording of the sentences generated by the AI ​​model based on the user's emotional state as recognized by the emotion engine. For example, if the user is excited, the language and style will be changed to a calmer one, and if the user is anxious, the language and style will be changed to a more reassuring one.

[0770] Real-time feedback

[0771] The emotion engine provides real-time feedback to users that reflects their emotions, allowing them to check whether their emotional state is reflected in their writing and make appropriate corrections.

[0772] 5. User Interface Module

[0773] View and edit drafts

[0774] The terminal displays the draft received from the server on the user interface. The user can check the displayed draft and make any necessary corrections or additions. The user can edit in real time.

[0775] Sending a message

[0776] After the user has completed the modifications, the final message is sent by clicking the "Send" button. The terminal will then construct the final message and process it for sending.

[0777] Specific examples

[0778] For example, consider a case where a user wants to send an important business email under emotional circumstances. The user logs in to the system, collects past business emails, and sends them to the server. The server learns from these and understands the user's writing style and suggested formats. When the user composes a new business email, the system recognizes the user's emotional state and generates a draft like the one below.

[0779] Dear [Recipient Name],

[0780] Thank you for your continued support.

[0781] This is your [username] at [your company name].

[0782] Below we provide additional information regarding the new project we proposed the other day.

[0783] ...

[0784] If you have any questions, please feel free to contact us.

[0785] Sincerely,

[0786] [User Name]

[0787] If the user is in a state of agitation, the system will adjust the tone of the sentence to a calmer and clearer tone, allowing the user to efficiently compose high-quality messages and ultimately achieve effective communication.

[0788] The above is a specific embodiment for carrying out the present invention. This system significantly reduces the effort required for writing text and enables natural and effective communication that takes into account the user's emotional state.

[0789] The processing flow will be explained below.

[0790] Step 1:

[0791] A user logs in to the system and links their email account or chat account to the system. The user then enters authentication information to allow the linking.

[0792] Step 2:

[0793] The device retrieves past email and chat history from linked accounts. The device collects this data through an API and encrypts it for privacy reasons.

[0794] Step 3:

[0795] The device sends the encrypted data to the server, which receives it and stores it securely in a database. The data remains encrypted while it is stored.

[0796] Step 4:

[0797] The server retrieves the stored emails and chat history from the database and performs pre-processing, which includes cleaning and tokenization, to ensure the data is in a suitable format for analysis.

[0798] Step 5:

[0799] The server extracts features from the preprocessed data, including the user's vocabulary, punctuation patterns, and line-breaking habits.

[0800] Step 6:

[0801] The server uses the extracted features to train an AI model, using a machine learning algorithm (e.g., a Transformer model) to generate a model that matches the user's writing style. This process also includes model evaluation and optimization.

[0802] Step 7:

[0803] The device's built-in emotion engine recognizes the user's emotions through their typing speed, keyboard usage, and facial expression recognition.

[0804] Step 8:

[0805] The user opens a new message composition screen. The device sends a request to the server to create a draft, including context information (subject and keywords).

[0806] Step 9:

[0807] The server uses a trained AI model to generate drafts in real time, which reflect the user's writing style and habits.

[0808] Step 10:

[0809] Based on the user's emotional state as recognized by the emotion engine, the server adjusts the writing style and language of the generated draft. For example, if the user is excited, the language and style will be changed to a calmer one.

[0810] Step 11:

[0811] The server generates and sends the adjusted draft to the terminal, which displays the draft on the user interface.

[0812] Step 12:

[0813] The user can check the displayed draft and make any necessary corrections or additions. The user can edit in real time.

[0814] Step 13:

[0815] The user clicks the send button on the completed message, and the terminal constructs the final message and processes it for sending.

[0816] Step 14:

[0817] The terminal sends the final message to the server, which then sends the received message to the destination via a mail server or chat server.

[0818] Example 2

[0819] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0820] Conventional writing support systems lack the ability to generate sentences that reflect the user's writing style and emotions, requiring the user to spend time manually correcting and adjusting them. Furthermore, sentences generated without considering the user's emotional state may result in a decline in the quality of communication. A system that solves these issues and generates high-quality sentences in real time that reflect the user's emotions while maintaining the user's writing style is needed.

[0821] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes means for collecting past communication data of a user, means for extracting features from the collected communication data, means for training an AI model based on the extracted features, means for generating a draft of a sentence in real time using the trained AI model, means for displaying the generated draft to the user, and means for recognizing the user's emotions and adjusting the style and wording of the generated sentence based on the user's emotional state. This enables real-time generation of high-quality sentences that reflect the user's style and emotions.

[0822] A "user" is an entity that uses this system to create, manage, and send communication data such as emails and chats.

[0823] "Communication data" refers to the content of emails and chats that a user has sent and received in the past.

[0824] "Features" are analytical units extracted from communication data, such as a user's unique language usage, punctuation patterns, and line break habits.

[0825] An "AI model" is an artificial intelligence algorithm trained based on features, which generates and predicts sentences.

[0826] The "emotion engine" is a function that recognizes the user's emotional state and adjusts the style and wording of the generated text according to that emotional state.

[0827] "Style adjustment" refers to changing the tone and expression of a sentence based on the user's emotional state and characteristics.

[0828] An "interface" is a screen or operating tool that allows a user to interact with the system, and provides functions for displaying, correcting, and sending drafts of text.

[0829] "Real-time" means responding immediately to user operations and immediately reflecting any results or changes that occur.

[0830] "Emotion recognition" refers to the technology or process of determining a user's emotional state based on their typing speed, keyboard usage, facial expressions, etc.

[0831] This invention is a system that collects a user's past communication data, extracts features, and generates documents using AI technology. It also includes an emotion engine that recognizes the user's emotions and adjusts the style and wording of the generated text accordingly. This system improves the efficiency of email and chat text creation and generates consistent, high-quality text.

[0832] User data integration

[0833] The user logs in to the system and links their email and chat accounts to the system. The user authenticates by entering their email address and password. Once the linking is complete, the device collects past email and chat history from these accounts via API and sends it to the server via encrypted communication.

[0834] Data Preprocessing

[0835] The server preprocesses the received communication data, cleansing it to remove noise and HTML tags, and then tokenizes it, breaking down sentences into words and converting them into a format suitable for analysis.

[0836] Feature extraction

[0837] The server extracts features from the preprocessed data, using natural language processing techniques to identify features such as the user's unique vocabulary, punctuation patterns, and line break habits.

[0838] Training an AI model

[0839] The server uses the extracted features to train an AI model, then uses a Transformer model as a machine learning algorithm to evaluate and optimize the model. This process generates the optimal model that best suits the user's writing style.

[0840] Draft Generation

[0841] When a user accesses the new message creation screen, the device sends a request to the server to generate a draft. The server uses a trained AI model to generate a draft in real time and sends it to the device. The generated draft reflects the user's writing style and habits.

[0842] Emotion recognition

[0843] The device's built-in emotion engine recognizes the user's emotions by analyzing their typing speed, keystroke patterns, and even facial expressions using a camera.

[0844] Emotion-based writing style adjustment

[0845] The emotion engine adjusts the style and language of the sentences generated by the AI ​​model based on the user's emotional state. For example, if the user is excited, the tone of the sentence will be changed to a calmer tone.

[0846] Real-time feedback

[0847] The device provides real-time feedback reflecting the user's emotions recognized by the emotion engine, allowing the user to see how their current emotional state is affecting their writing and make appropriate corrections.

[0848] View and edit drafts

[0849] The terminal displays the draft received from the server on the user interface. The user can check the displayed draft and make any necessary corrections or additions. The user can edit in real time and the changes are reflected immediately.

[0850] Sending a message

[0851] Once the user has finished editing the message, they click the send button to send the final message. The device then constructs the final message and sends it to the destination via the server in an encrypted connection.

[0852] Specific examples

[0853] For example, consider a case where a user wants to send an important business email under emotional circumstances. The user logs in to the system, collects past business emails, and sends them to the server. The server learns from these and understands the user's writing style and suggested formats. When the user composes a new business email, the system recognizes the user's emotional state and generates a draft like the one below.

[0854] Dear [Recipient Name],

[0855] Thank you for your continued support.

[0856] This is your [username] at [your company name].

[0857] Below we provide additional information regarding the new project we proposed the other day.

[0858] ...

[0859] If you have any questions, please feel free to contact us.

[0860] Sincerely,

[0861] [User Name]

[0862] If the user is in a state of agitation, the system can adjust the tone of the sentence to a calmer and clearer tone, allowing the user to efficiently compose high-quality messages and ultimately achieve effective communication.

[0863] The above is a specific embodiment for carrying out the present invention. This system significantly reduces the effort required for writing text and enables natural and effective communication that takes into account the user's emotional state.

[0864] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0865] Divide the program's processing flow into processing steps

[0866] Step 1: Linking user data

[0867] Step 2: Preprocessing the data

[0868] Step 3: Feature extraction

[0869] Step 4: Training the AI ​​model

[0870] Step 5: Draft generation

[0871] Step 6: Recognize emotions

[0872] Step 7: Adjust your writing style based on emotion

[0873] Step 8: Real-time feedback

[0874] Step 9: View and edit your draft

[0875] Step 10: Sending a message

[0876] Specific explanation of each step

[0877] Step 1: Linking user data

[0878] A user logs in to the system.

[0879] The user enters an email address and password to log in to the system and is authenticated.

[0880] Input: Email address, password

[0881] Output: User's authentication token

[0882] The device is linked to the user's email and chat accounts.

[0883] With the user's permission, the device accesses email and chat accounts through an API.

[0884] Input: Authentication token, account information

[0885] Output: Access rights

[0886] The collected data is sent to a server.

[0887] The device encrypts the collected past emails and chat history and sends it to the server.

[0888] Input: Past communication data

[0889] Output: Encrypted data sent to the server

[0890] Step 2: Preprocessing the data

[0891] The server preprocesses the data received.

[0892] The server cleans the data it receives, removing noise and HTML tags, and then tokenizes it to convert it into a format that is easier to analyze.

[0893] Input: Received communication data

[0894] Output: Preprocessed data

[0895] Step 3: Feature extraction

[0896] The server extracts features from the preprocessed data.

[0897] The server uses natural language processing technology to extract features such as the user's unique vocabulary, punctuation patterns, and line break habits.

[0898] Input: Preprocessed data

[0899] Output: Extracted features

[0900] Step 4: Training the AI ​​model

[0901] The server uses the extracted features to train an AI model.

[0902] The server uses the Transformer model as a machine learning algorithm to evaluate and optimize the model.

[0903] Input: extracted features

[0904] Output: A trained AI model

[0905] Step 5: Draft generation

[0906] The user accesses a new message composition screen.

[0907] The user logs into the system interface using a terminal and opens a message composition screen.

[0908] Input: User interface access

[0909] Output: Message creation screen

[0910] The terminal sends a request to the server to generate a draft.

[0911] The device sends a request to the server to generate a draft using the trained AI model.

[0912] Input: Draft generation request

[0913] Output: A draft creation request is sent to the server

[0914] The server sends the generated draft to the device.

[0915] The server returns the draft generated by the AI ​​model to the device.

[0916] Input: A trained AI model

[0917] Output: Draft document

[0918] Step 6: Recognize emotions

[0919] The emotion engine installed in the device recognizes the user's emotions.

[0920] The device analyzes keystroke speed, typing patterns and facial expressions using a camera to determine the user's emotions.

[0921] Input: key input data, keystroke patterns, facial expression data

[0922] Output: Recognized user emotion data

[0923] Step 7: Adjust your writing style based on emotion

[0924] The emotion engine adjusts writing style and language based on the emotional state it recognizes.

[0925] The server uses AI models to, for example, change the tone of the text to a calmer tone if the user is in an agitated state.

[0926] Input: Recognized emotion data, draft text

[0927] Output: Adjusted sentence

[0928] Step 8: Real-time feedback

[0929] The device provides real-time feedback reflecting the user's emotions recognized by the emotion engine.

[0930] The device displays emotional feedback on the screen, allowing users to see how their current emotional state is affecting their writing.

[0931] Input: Recognized emotion data

[0932] Output: Emotional feedback to the user

[0933] Step 9: View and edit your draft

[0934] The terminal displays the draft received from the server on the user interface.

[0935] The device provides an interface that allows you to edit the displayed draft as it is.

[0936] Input: Draft text

[0937] Output: Editable interface

[0938] The user checks the displayed draft and makes any necessary corrections or additions.

[0939] Users can edit in real time and see changes reflected immediately.

[0940] Input: Draft text

[0941] Output: Final text with corrections and additions

[0942] Step 10: Sending a message

[0943] The user clicks the send button to send the message.

[0944] After the user has completed the modifications, the final message is sent by clicking the send button.

[0945] Input: Final text with corrections and additions

[0946] Output: Send request

[0947] The terminal constructs the final message and sends it to the destination via the server.

[0948] The device encrypts the message according to the destination protocol and sends it to the intended recipient via the server.

[0949] Input: Send request, destination information

[0950] Output: Message sent

[0951] The above are the specific processing steps and their respective operations of this system, which make the writing process more efficient and realize natural communication that reflects the user's emotional state.

[0952] (Application example 2)

[0953] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0954] Customer support plays an important role in modern electronic payment services. However, while prompt and appropriate responses are required for user inquiries and troubleshooting, they often require a great deal of time and effort. Providing an efficient response is particularly difficult when it is necessary to take into account the user's emotional state. The purpose of this invention is to solve these problems and automatically generate prompt and emotionally sensitive responses.

[0955] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for collecting past communication data of a user, means for extracting features from the collected communication data, means for training an AI model based on the extracted features, means for generating drafts of sentences in real time using the trained AI model, means for displaying the generated drafts to the user, means for grasping the user's emotional state, and means for adjusting the style and wording of the generated sentences based on the grasped emotional state. This enables high-quality response sentences to be generated quickly and with consideration for emotions.

[0956] "User's past communication data" refers to the communication history of the user, such as emails and chats that the user has sent and received in the past.

[0957] "Features" are important attributes or patterns extracted from data to analyze the information.

[0958] An "AI model" is a computational model that uses artificial intelligence technology to learn from data and make predictions and classifications.

[0959] "Generating a draft of a text in real time" means creating a first draft of a text based on data that is instantly input.

[0960] "Displaying the generated draft to the user" means displaying the first draft generated by AI on the user's device screen.

[0961] Understanding the "user's emotional state" means observing and analyzing the user's current mental and emotional state.

[0962] "Adjusting style and language" means changing the way you write and express things appropriately.

[0963] An "interface" refers to the operation screen and input device that allow the user to interact with the system.

[0964] "Other people's communication data" refers to communication history such as emails and chats sent and received in the past by individuals other than the user.

[0965] "Imitating a style" means reproducing the writing style or expression of a particular person.

[0966] This invention is a system for automatically generating optimal responses to user inquiries and problems in customer support for electronic payment services. This system takes into consideration the user's emotional state and can generate quick, high-quality responses.

[0967] Overall system overview

[0968] The system primarily uses the following hardware and software components:

[0969] server

[0970] User's device (PC or smartphone)

[0971] Natural language processing libraries (e.g. spaCy or NLTK)

[0972] Transformer models (e.g., BERT and GPT-3)

[0973] Emotion recognition software (e.g., face-api.js)

[0974] Data Collection Module

[0975] When a user logs in to the service, the system collects the user's past communication data, including past emails and chat history. This data is securely transmitted to the server using SSL / TLS encryption protocols.

[0976] Model Learning Module

[0977] The server preprocesses the incoming communication data. This includes cleaning and tokenizing the data using an automated NLP library (e.g., spaCy or NLTK). Next, it extracts features from the data (e.g., wording and punctuation patterns). These features are used to retrain a Transformer model (e.g., BERT or GPT-3) to match the writing style of a specific user.

[0978] Emotion Recognition Module

[0979] The user's device monitors the user's typing speed and patterns in real time, and in some cases analyzes the user's emotional state using facial recognition technology (e.g., face-api.js) using a webcam. This information is used to understand the user's current emotional state (e.g., impatience, anxiety).

[0980] Text generation and style adjustment module

[0981] The trained AI model generates real-time draft responses to user queries, which are then sent from the server to the user's device, while adjusting the writing style and wording based on data obtained from the emotion recognition module.

[0982] User Interface Module

[0983] The generated draft is displayed on the user's screen, and the user can make corrections or additions in real time. The final confirmed message is sent by clicking the "Send" button.

[0984] Examples of specific examples and prompts

[0985] For example, if a user reports a problem with a "payment not being posted," the following response might be generated:

[0986] Dear (Username),

[0987] Thank you for your help. This is (Name of person in charge) from the customer support department of the electronic payment service.

[0988] We will look into the issue of your reported payment not being reflected as soon as possible. We apologize for any inconvenience this may cause.

[0989] Could you please provide the following information:

[0990] Transaction ID at the time of payment

[0991] Payment date and time

[0992] Payment amount

[0993] We will respond promptly based on the information you provide.

[0994] Thank you for your cooperation.

[0995] Sincerely,

[0996] (Name of person in charge)

[0997] If the user is in a state of agitation, the tone of the text will be adjusted to a calmer tone.

[0998] Example prompt sentence:

[0999] "Generate a response for a situation where a user is reporting difficulty with a payment and the user is in a panicked state. Consider the user's past inquiry history."

[1000] This system will significantly improve the efficiency and quality of customer support for electronic payment services.

[1001] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1002] Step 1:

[1003] A user logs in to the service. At this time, the user links their email account and chat account to the system. After linking, the user's device collects past email and chat history from these accounts. The collected data is encrypted and sent to the server. The input is the user's login information and communication data, and the output is the transmission of encrypted data.

[1004] Step 2:

[1005] The server preprocesses the communication data it receives. Specifically, it cleans and tokenizes the data using a natural language processing library (such as spaCy or NLTK). During this process, it removes unnecessary symbols and spaces and splits the data into words. The input is the decrypted communication data, and the output is the preprocessed, clean data.

[1006] Step 3:

[1007] The server extracts features from the preprocessed data. Features include the user's vocabulary, punctuation patterns, and line break habits. A feature extraction algorithm is used for extraction. The input is the preprocessed clean data, and the output is the features.

[1008] Step 4:

[1009] The server uses the extracted features to train an AI model. A Transformer model (e.g., BERT or GPT-3) is retrained with the specific user's data. This process also includes model evaluation and optimization. The input is the features, and the output is the trained AI model.

[1010] Step 5:

[1011] The user's device monitors the user's input speed and patterns in real time to grasp their emotional state. In some cases, a webcam is used to recognize facial expressions and analyze the user's emotions. The input is the user's input behavior and facial image, and the output is data on the user's emotional state.

[1012] Step 6:

[1013] When a user accesses the new message composition screen, the device sends a request to the server to generate a draft. The server uses a trained AI model to generate the draft in real time. The draft is based on the user's past communication data. The input is the user's request, and the output is the initial draft.

[1014] Step 7:

[1015] The server adjusts the writing style and language of the draft generated by the server based on data obtained from the emotion recognition module. For example, if the user is in a state of agitation, the writing style will be made more calm. The input is the initial draft and emotion data, and the output is the adjusted draft.

[1016] Step 8:

[1017] The generated draft is displayed on the user's device. The user can review the displayed draft and make corrections or additions as needed. This correction is done in real time. The input is the adjusted draft, and the output is the final message as modified by the user.

[1018] Step 9:

[1019] After the user completes the modifications, they click the "Send" button to send the final message. The terminal constructs the final message and processes it for sending. The input is the final message modified by the user, and the output is the sent message.

[1020] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[1021] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1022] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the smart glasses 214.

[1023] [Third embodiment]

[1024] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[1025] 5, the data processing system 310 includes the data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[1026] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1027] The headset type terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the display 343 are also connected to the bus 52.

[1028] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[1029] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1030] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1031] Fig. 6 shows an example of the main functions of the data processing device 12 and the headset type terminal 314. As shown in Fig. 6, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1032] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1033] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1034] In the headset type terminal 314, a reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1035] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the headset type terminal 314 will be referred to as the "terminal."

[1036] This invention is a system that collects users' past communication data, extracts features, and generates documents using AI technology. This system is designed to improve the efficiency of email and chat document creation and generate consistent, high-quality documents.

[1037] 1. Data Collection Module

[1038] User data integration

[1039] Users log in to the system and link their email and chat accounts. After linking, the user's device collects past email and chat history from these accounts and sends it to the server. Data transmission is encrypted to protect privacy.

[1040] 2. Model Learning Module

[1041] Data Preprocessing

[1042] The server preprocesses the communication data it receives. Specifically, it cleans and tokenizes the text to extract features such as the user's vocabulary, punctuation, and line break habits.

[1043] Training an AI model

[1044] The features are used to train an AI model. The server uses a machine learning algorithm (e.g., a Transformer model) to build a model that generates sentences that match the user's writing style.

[1045] 3. Sentence Generation Module

[1046] Draft Generation

[1047] When a user accesses the new message creation screen, the device sends a request to the server to generate a draft. The server uses a trained AI model to generate a draft in real time and sends it to the device. The generated draft reflects the user's writing style and habits.

[1048] 4. User Interface Module

[1049] View and edit drafts

[1050] The terminal displays the draft received from the server on the user interface. The user can check the displayed draft and make corrections or additions as necessary. The corrected message is completed on the terminal without being sent to the server again.

[1051] Sending a message

[1052] After the user has completed the edits, the final message is sent by clicking the "Send" button. The terminal then performs the process to send the final message.

[1053] Specific examples

[1054] For example, consider a situation where a business user wants to send an email proposing a new project. The user uses the system and uploads hundreds of past proposal emails to the system. The system learns from these emails and understands the user's writing style and proposal format. When the user creates a new proposal email, the system generates a draft like the one below.

[1055] Dear [Recipient Name],

[1056] Thank you for your continued support.

[1057] This is your [username] at [your company name].

[1058] Below we provide additional information regarding the new project we proposed the other day.

[1059] ...

[1060] If you have any questions, please feel free to contact us.

[1061] Sincerely,

[1062] [User Name]

[1063] Users can review this draft, add specific project details and dates, and then send it, allowing them to efficiently create consistent, high-quality messages.

[1064] The above is a specific embodiment for carrying out the present invention. This system significantly reduces the effort required for writing text and enables natural and effective communication.

[1065] The processing flow will be explained below.

[1066] Step 1:

[1067] A user logs in to the system and links their email account or chat account to the system. The user then enters authentication information to allow the linking.

[1068] Step 2:

[1069] The device retrieves past email and chat history from linked accounts. The device collects this data through an API and encrypts it for privacy reasons.

[1070] Step 3:

[1071] The device sends the encrypted data to the server, which receives it and stores it securely in a database. The data remains encrypted while it is stored.

[1072] Step 4:

[1073] The server retrieves the stored emails and chat history from the database and performs pre-processing, which includes cleaning and tokenization, to ensure the data is in a suitable format for analysis.

[1074] Step 5:

[1075] The server extracts features from the preprocessed data, including the user's vocabulary, punctuation patterns, and line-breaking habits.

[1076] Step 6:

[1077] The server uses the extracted features to train an AI model, using a machine learning algorithm (e.g., a Transformer model) to generate a model that matches the user's writing style. This process also includes model evaluation and optimization.

[1078] Step 7:

[1079] The user opens a new message composition screen. The device sends a request to the server to create a draft, including context information (subject and keywords).

[1080] Step 8:

[1081] The server uses a trained AI model to generate drafts in real time, which reflect the user's writing style and habits.

[1082] Step 9:

[1083] The server sends the generated draft to the terminal, which displays the draft on the user interface.

[1084] Step 10:

[1085] The user can check the displayed draft and make any necessary corrections or additions. The user can make these edits in real time using the device interface.

[1086] Step 11:

[1087] The user clicks the send button on the completed message, and the terminal constructs the final message and processes it for sending.

[1088] Step 12:

[1089] The terminal sends the final message to the server, which then sends the received message to the destination via a mail server or chat server.

[1090] Example 1

[1091] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1092] In today's business environment, users are required to maintain fast, consistent, and high-quality communications. However, manually creating documents is time-consuming and labor-intensive, making it difficult, especially when large volumes of communication are required. Privacy protection and data security are also important issues. Therefore, there is a need for a system that can effectively utilize users' past communication data and generate high-quality documents while safely managing the data.

[1093] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1094] In this invention, the server includes means for collecting past communication data of users, means for extracting features from the collected communication data, means for training an AI model based on the extracted features, means for generating drafts of sentences in real time using the trained AI model, means for displaying the generated drafts to the user, and means for protecting data using encryption technology, thereby enabling users to generate high-quality sentences quickly and consistently, while also ensuring privacy protection and data security.

[1095] "User's past communication data" is a collection of digital data related to communications such as emails and chat messages that the user has sent and received in the past.

[1096] "Features" refer to specific attributes and patterns extracted from collected communication data, including, for example, a user's vocabulary, punctuation, and line break habits.

[1097] An "AI model" is an artificial intelligence framework that has been trained using machine learning algorithms to perform a specific task (in this case, sentence generation).

[1098] "Training" refers to the process of having an AI model learn from collected data and acquire the ability to accomplish a given task.

[1099] A "draft" is an initial version of a message that a user is about to create, and is a basic sentence generated by an AI model.

[1100] "Encryption technology" refers to a technical method used to prevent data from being deciphered by third parties, and is used particularly to increase the security of communications.

[1101] A "user interface" is a display screen or operation panel that allows a user to interact with a system.

[1102] This invention is a system that collects users' past communication data, extracts features, and generates documents using AI technology. This system is designed to improve the efficiency of email and chat document creation and generate consistent, high-quality documents.

[1103] Data Collection Module

[1104] The user logs in to the system and links their email and chat accounts. The device collects past email and chat history from the linked accounts, encrypts it, and sends it to the server. AES encryption technology is used for encryption. The server securely stores the received data.

[1105] Model Learning Module

[1106] The server preprocesses the received communication data by cleaning and tokenizing it, and extracting features such as the user's vocabulary, punctuation, and line-breaking habits. To do this, it uses an NLP (Natural Language Processing) library (e.g., SpaCy or NLTK).

[1107] The server then trains an AI model based on the features. It uses a transformer model (e.g., BERT or GPT) and builds the model on a GPU cloud (e.g., NVIDIA GPU cloud). After the training process, the server creates a sentence generation model that matches the user's writing style.

[1108] Sentence Generation Module

[1109] When a user accesses the new message creation screen, the device sends a request to the server to generate a draft. The server uses a trained AI model to generate a draft in real time and sends it to the device. The generated draft reflects the user's writing style and habits.

[1110] User Interface Module

[1111] The terminal displays the draft received from the server on the user interface. The user can check the displayed draft and make corrections or additions as necessary. In this case, the corrected message is completed on the terminal without being resent to the server.

[1112] Finally, when the user clicks the "send" button, the device processes the final message to send it, allowing users to efficiently create consistent, high-quality messages.

[1113] Specific examples

[1114] For example, consider a situation where a business user wants to send an email proposing a new project. The user uses the system and uploads hundreds of past proposal emails to the system. The system learns from these emails and understands the user's writing style and proposal format. When the user creates a new proposal email, the system generates a draft like the one below.

[1115] Dear [Recipient Name],

[1116] Thank you for your continued support.

[1117] This is your [username] at [your company name].

[1118] Below we provide additional information regarding the new project we proposed the other day.

[1119] ...

[1120] If you have any questions, please feel free to contact us.

[1121] Sincerely,

[1122] [User Name]

[1123] Users can review this draft, add specific project details and dates, and then send it, allowing them to efficiently create consistent, high-quality messages.

[1124] The above is a specific embodiment for carrying out the present invention. This system significantly reduces the effort required for writing text and enables natural and effective communication.

[1125] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1126] Step 1:

[1127] User login and account linking

[1128] 1. A user logs into the system.

[1129] Input: User ID and password

[1130] Output: Authentication result (success / failure)

[1131] How it works: The user enters their ID and password on the dedicated login screen and clicks the login button. The server performs authentication, and if login is successful, their personal page is displayed.

[1132] 2. Users link their email and chat accounts to the system.

[1133] Input: Account information (email address, API key, etc.)

[1134] Output: Notification of integration success / failure

[1135] How it works: The user selects "Account Linking" from the settings screen and enters their account information. When they click the link button, the device will link their account and display a notification when the linking is complete.

[1136] Step 2:

[1137] Data Collection and Encryption

[1138] 1. Collect past email and chat history from the account linked to the device.

[1139] Input: Linked account information

[1140] Output: Collected communication data

[1141] How it works: The device uses APIs and other protocols to collect past email and chat history from your account. The collected data is temporarily stored on the device.

[1142] 2. The collected data is encrypted and sent to the server.

[1143] Input: communication data

[1144] Output: Encrypted data

[1145] How it works: Your device encrypts data using AES encryption and sends it to the server, which then decrypts it and stores it securely.

[1146] Step 3:

[1147] Data Preprocessing

[1148] 1. The server preprocesses the data it receives.

[1149] Input: Received encrypted data

[1150] Output: Preprocessed text data

[1151] How it works: The server decrypts the encrypted data and cleans it, specifically removing unnecessary metadata and duplicate messages.

[1152] 2. The server tokenizes the text and extracts features.

[1153] Input: Cleaned text data

[1154] Output: Tokenized data and features

[1155] How it works: The server uses an NLP (natural language processing) library (e.g., SpaCy or NLTK) to tokenize the text and extract features such as the user's vocabulary, punctuation, and line-breaking habits.

[1156] Step 4:

[1157] Training an AI model

[1158] 1. The server trains an AI model based on the features.

[1159] Input: Feature data

[1160] Output: A trained AI model

[1161] How it works: The server uses a Transformer model (e.g., BERT or GPT) to train an AI model on a GPU cloud (e.g., NVIDIA GPU cloud).

[1162] Step 5:

[1163] Real-time draft generation

[1164] 1. The user accesses the new message composition screen.

[1165] Input: User action (clicking New)

[1166] Output: Draft creation request

[1167] Behavior: The user clicks the "Create New" button on the message composition screen. The device sends a JSON-formatted draft creation request to the server.

[1168] 2. The server generates drafts in real time using a trained AI model.

[1169] Input: Draft generation request

[1170] Output: Generated draft

[1171] How it works: The server processes the draft generation request, generates the draft using the trained AI model, and returns the generated text in JSON format to the device.

[1172] Step 6:

[1173] Check and correct via user interface

[1174] 1. The terminal displays the received draft on the user interface.

[1175] Input: Generated draft data

[1176] Output: Draft text displayed

[1177] Operation: The device parses the received JSON data and displays it in the user interface via a text editor.

[1178] 2. The user reviews the draft and makes corrections or additions as necessary.

[1179] Input: Draft text displayed

[1180] Output: Corrected text

[1181] How it works: A user edits a draft in a text editor and then reviews the completed message.

[1182] Step 7:

[1183] Sending the final message

[1184] 1. The user clicks the "Submit" button.

[1185] Input: User action (click submit)

[1186] Output: Start of transmission process

[1187] Behavior: When the user clicks the "Send" button, the device processes the final message to send it.

[1188] 2. The device sends a message.

[1189] Input: Final modified message

[1190] Output: Message sent

[1191] What it does: Your device sends a message to the other person via email or chat account.

[1192] The above are the specific processing steps of the system program, which enables users to create and send messages efficiently, consistently, and with high quality.

[1193] (Application example 1)

[1194] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1195] Customer support in food delivery services requires quick and consistent responses, which places a heavy burden on customer support agents. Furthermore, when different agents respond, inconsistent writing styles and response policies can lead to lower customer satisfaction. To solve this problem, an efficient and consistent message generation system is needed.

[1196] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1197] In this invention, the server includes means for collecting past communication data of a user, means for extracting features from the collected communication data, means for training an AI model based on the extracted features, means for generating message drafts in real time using the trained AI model, means for displaying the generated drafts to the user, and means for automating customer support in a food delivery service, thereby enabling customer support personnel to efficiently generate consistent, high-quality messages and improving customer satisfaction.

[1198] "User" refers to the customer support staff of the food delivery service.

[1199] "Communication data" refers to records of chat and email exchanges between users and customers that have been collected in the past.

[1200] "Features" refer to the user's vocabulary and structural features extracted from communication data.

[1201] "AI model" refers to an artificial intelligence model that is trained based on features and learns the user's writing style and response patterns.

[1202] "Message Draft" refers to a draft response to a customer that is generated in real time by a trained AI model.

[1203] The "means for displaying" refers to a function for displaying the generated draft of the message on the user's terminal.

[1204] "Food delivery service" refers to all services that involve ordering and delivering meals.

[1205] "Customer response automation" refers to the ability for users to use AI models to automatically provide appropriate responses to customer inquiries and complaints.

[1206] The present invention provides a system for improving the efficiency of customer service in food delivery services and generating consistent, high-quality messages. Specific embodiments of this system are described below.

[1207] 1. Data Collection Module

[1208] The server collects users' past communication data. This communication data is a record of chats and emails exchanged between customer support representatives and customers. The data is encrypted and appropriate measures are taken to protect privacy. The specific software used is an API that uses the requests library to retrieve data from the server.

[1209] 2. Model Learning Module

[1210] The server preprocesses the collected communication data and extracts features, such as stylistic and structural characteristics, from the text. This involves text processing such as tokenization and cleaning. An AI model is then trained using specific machine learning algorithms, such as AutoTokenizer and AutoModelForSeq2SeqLM from the transformers library. This training builds a model that matches the user's writing style.

[1211] 3. Sentence Generation Module

[1212] The server uses a trained AI model to generate message drafts in real time. When a user enters a prompt, the AI ​​model generates an appropriate reply message draft and sends the result to the user's device. This allows for fast, high-quality messages.

[1213] 4. User Interface Module

[1214] The terminal displays the generated draft message to the user. The user can check the displayed draft and make corrections as necessary. The corrected message can then be sent to the customer as is. This allows the user to respond consistently without any hassle.

[1215] Specific examples

[1216] For example, consider the following prompt text entered by a customer support representative:

[1217] (prompt statement)

[1218] "Complaint handling: If a customer reports that there was a foreign object in the food they ordered, how would you respond?"

[1219] Based on this prompt, the AI ​​model generates a draft message like this:

[1220] (Generated Draft)

[1221] "Dear customers,

[1222] Thank you for using our food delivery service.

[1223] We deeply apologize for the report that a foreign object was found in the food you received.

[1224] We will investigate the matter immediately and respond promptly, so please wait a moment.

[1225] We will also give you a special discount on your next order.

[1226] Thank you for your continued support.

[1227] Sincerely,

[1228] Customer Support Representative Name

[1229] The system of the present invention enables customer support personnel to provide efficient and consistent responses, thereby improving customer satisfaction.

[1230] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1231] Step 1: User logs into the system

[1232] The user logs in to the system and links to their food delivery service account. At this time, they enter authentication information such as their ID and password. The system uses the entered authentication information to authenticate the user and allow them to log in. This makes it possible to access the user's past communication data.

[1233] Step 2: Collecting communication data

[1234] The server collects the user's past communication data through the API. Specifically, it uses the requests library to obtain the communication data using the API token and user ID. This data includes chat history with customers and email content. It receives the API token and user ID as input and obtains the communication data in JSON format as output.

[1235] Step 3: Preprocessing the data

[1236] The server preprocesses the collected communication data. This preprocessing includes cleaning and tokenizing the communication data, and uses AutoTokenizer from the transformers library. It receives JSON-formatted communication data as input and obtains clean text data as output. This makes it easier to extract noise-free features.

[1237] Step 4: Feature extraction

[1238] The server extracts features from the preprocessed communication data. Features refer to the user's writing style and structural characteristics, such as word frequency and sentence structure. It receives clean text data as input and obtains a feature vector as output. This feature vector is used for subsequent AI model training.

[1239] Step 5: Training the AI ​​model

[1240] The server trains an AI model based on the features. It uses AutoModelForSeq2SeqLM from the transformers library, receives the feature vector as input, and obtains a trained AI model as output. The trained AI model generates sentences that match the user's writing style.

[1241] Step 6: Enter a prompt statement

[1242] When a user creates a new message, they input a prompt text. For example, they can input a specific prompt text such as "Complaint handling: If a customer reports that a foreign object was found in the food they ordered, how would you respond?" The prompt text is received as input and sent to the system.

[1243] Step 7: Draft your message

[1244] The server generates a message draft in real time based on the input prompt. It uses the transformers library to process the prompt with an AI model. It takes the prompt as input and generates a message draft as output. This draft is adapted to the user's writing style.

[1245] Step 8: View and edit your draft

[1246] The terminal displays the generated message draft to the user. The user can review the displayed draft and make corrections as necessary. The corrected message is saved immediately and is not sent to the server again. It receives the message draft as input and displays it to the user as output.

[1247] Step 9: Sending a message

[1248] After the user completes the modification, the final message is sent. The terminal processes the final message and sends a quick reply to the customer. It receives the modified message as input and sends the final message to the customer as output.

[1249] This process allows customer support representatives at food delivery services to efficiently generate consistent, high-quality messages, improving customer satisfaction.

[1250] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1251] This invention is a system that collects a user's past communication data, extracts features, and generates documents using AI technology. Furthermore, this invention incorporates an emotion engine that recognizes the user's emotions and adjusts the style and wording of the generated text based on those emotions. This system improves the efficiency of email and chat text creation and generates consistent, high-quality text.

[1252] 1. Data Collection Module

[1253] User data integration

[1254] Users log in to the system and link their email and chat accounts. After linking, the user's device collects past email and chat history from these accounts and sends it to the server. Data transmission is encrypted to protect privacy.

[1255] 2. Model Learning Module

[1256] Data Preprocessing

[1257] The server pre-processes the received communication data, which includes cleaning and tokenization, to make the data suitable for analysis.

[1258] Feature extraction

[1259] The server extracts features from the preprocessed data, including the user's vocabulary, punctuation patterns, and line-breaking habits.

[1260] Training an AI model

[1261] The server uses the extracted features to train an AI model, using a machine learning algorithm (e.g., a Transformer model) to generate a model that matches the user's writing style. This process also includes model evaluation and optimization.

[1262] 3. Sentence Generation Module

[1263] Draft Generation

[1264] When a user accesses the new message creation screen, the device sends a request to the server to generate a draft. The server uses a trained AI model to generate a draft in real time and sends it to the device. The generated draft reflects the user's writing style and habits.

[1265] 4. Emotion Engine

[1266] Emotion recognition

[1267] The device's built-in emotion engine recognizes the user's emotions through, for example, the user's typing speed, keyboard usage, and facial expression recognition.

[1268] Emotion-based writing style adjustment

[1269] The emotion engine adjusts the style and wording of the sentences generated by the AI ​​model based on the user's emotional state as recognized by the emotion engine. For example, if the user is excited, the language and style will be changed to a calmer one, and if the user is anxious, the language and style will be changed to a more reassuring one.

[1270] Real-time feedback

[1271] The emotion engine provides real-time feedback to users that reflects their emotions, allowing them to check whether their emotional state is reflected in their writing and make appropriate corrections.

[1272] 5. User Interface Module

[1273] View and edit drafts

[1274] The terminal displays the draft received from the server on the user interface. The user can check the displayed draft and make any necessary corrections or additions. The user can edit in real time.

[1275] Sending a message

[1276] After the user has completed the modifications, the final message is sent by clicking the "Send" button. The terminal will then construct the final message and process it for sending.

[1277] Specific examples

[1278] For example, consider a case where a user wants to send an important business email under emotional circumstances. The user logs in to the system, collects past business emails, and sends them to the server. The server learns from these and understands the user's writing style and suggested formats. When the user composes a new business email, the system recognizes the user's emotional state and generates a draft like the one below.

[1279] Dear [Recipient Name],

[1280] Thank you for your continued support.

[1281] This is your [username] at [your company name].

[1282] Below we provide additional information regarding the new project we proposed the other day.

[1283] ...

[1284] If you have any questions, please feel free to contact us.

[1285] Sincerely,

[1286] [User Name]

[1287] If the user is in a state of agitation, the system will adjust the tone of the sentence to a calmer and clearer tone, allowing the user to efficiently compose high-quality messages and ultimately achieve effective communication.

[1288] The above is a specific embodiment for carrying out the present invention. This system significantly reduces the effort required for writing text and enables natural and effective communication that takes into account the user's emotional state.

[1289] The processing flow will be explained below.

[1290] Step 1:

[1291] A user logs in to the system and links their email account or chat account to the system. The user then enters authentication information to allow the linking.

[1292] Step 2:

[1293] The device retrieves past email and chat history from linked accounts. The device collects this data through an API and encrypts it for privacy reasons.

[1294] Step 3:

[1295] The device sends the encrypted data to the server, which receives it and stores it securely in a database. The data remains encrypted while it is stored.

[1296] Step 4:

[1297] The server retrieves the stored emails and chat history from the database and performs pre-processing, which includes cleaning and tokenization, to ensure the data is in a suitable format for analysis.

[1298] Step 5:

[1299] The server extracts features from the preprocessed data, including the user's vocabulary, punctuation patterns, and line-breaking habits.

[1300] Step 6:

[1301] The server uses the extracted features to train an AI model, using a machine learning algorithm (e.g., a Transformer model) to generate a model that matches the user's writing style. This process also includes model evaluation and optimization.

[1302] Step 7:

[1303] The device's built-in emotion engine recognizes the user's emotions through their typing speed, keyboard usage, and facial expression recognition.

[1304] Step 8:

[1305] The user opens a new message composition screen. The device sends a request to the server to create a draft, including context information (subject and keywords).

[1306] Step 9:

[1307] The server uses a trained AI model to generate drafts in real time, which reflect the user's writing style and habits.

[1308] Step 10:

[1309] Based on the user's emotional state as recognized by the emotion engine, the server adjusts the writing style and language of the generated draft. For example, if the user is excited, the language and style will be changed to a calmer one.

[1310] Step 11:

[1311] The server generates and sends the adjusted draft to the terminal, which displays the draft on the user interface.

[1312] Step 12:

[1313] The user can check the displayed draft and make any necessary corrections or additions. The user can edit in real time.

[1314] Step 13:

[1315] The user clicks the send button on the completed message, and the terminal constructs the final message and processes it for sending.

[1316] Step 14:

[1317] The terminal sends the final message to the server, which then sends the received message to the destination via a mail server or chat server.

[1318] Example 2

[1319] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1320] Conventional writing support systems lack the ability to generate sentences that reflect the user's writing style and emotions, requiring the user to spend time manually correcting and adjusting them. Furthermore, sentences generated without considering the user's emotional state may result in a decline in the quality of communication. A system that solves these issues and generates high-quality sentences in real time that reflect the user's emotions while maintaining the user's writing style is needed.

[1321] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes means for collecting past communication data of a user, means for extracting features from the collected communication data, means for training an AI model based on the extracted features, means for generating a draft of a sentence in real time using the trained AI model, means for displaying the generated draft to the user, and means for recognizing the user's emotions and adjusting the style and wording of the generated sentence based on the user's emotional state. This enables real-time generation of high-quality sentences that reflect the user's style and emotions.

[1322] A "user" is an entity that uses this system to create, manage, and send communication data such as emails and chats.

[1323] "Communication data" refers to the content of emails and chats that a user has sent and received in the past.

[1324] "Features" are analytical units extracted from communication data, such as a user's unique language usage, punctuation patterns, and line break habits.

[1325] An "AI model" is an artificial intelligence algorithm trained based on features, which generates and predicts sentences.

[1326] The "emotion engine" is a function that recognizes the user's emotional state and adjusts the style and wording of the generated text according to that emotional state.

[1327] "Style adjustment" refers to changing the tone and expression of a sentence based on the user's emotional state and characteristics.

[1328] An "interface" is a screen or operating tool that allows a user to interact with the system, and provides functions for displaying, correcting, and sending drafts of text.

[1329] "Real-time" means responding immediately to user operations and immediately reflecting any results or changes that occur.

[1330] "Emotion recognition" refers to the technology or process of determining a user's emotional state based on their typing speed, keyboard usage, facial expressions, etc.

[1331] This invention is a system that collects a user's past communication data, extracts features, and generates documents using AI technology. It also includes an emotion engine that recognizes the user's emotions and adjusts the style and wording of the generated text accordingly. This system improves the efficiency of email and chat text creation and generates consistent, high-quality text.

[1332] User data integration

[1333] The user logs in to the system and links their email and chat accounts to the system. The user authenticates by entering their email address and password. Once the linking is complete, the device collects past email and chat history from these accounts via API and sends it to the server via encrypted communication.

[1334] Data Preprocessing

[1335] The server preprocesses the received communication data, cleansing it to remove noise and HTML tags, and then tokenizes it, breaking down sentences into words and converting them into a format suitable for analysis.

[1336] Feature extraction

[1337] The server extracts features from the preprocessed data, using natural language processing techniques to identify features such as the user's unique vocabulary, punctuation patterns, and line break habits.

[1338] Training an AI model

[1339] The server uses the extracted features to train an AI model, then uses a Transformer model as a machine learning algorithm to evaluate and optimize the model. This process generates the optimal model that best suits the user's writing style.

[1340] Draft Generation

[1341] When a user accesses the new message creation screen, the device sends a request to the server to generate a draft. The server uses a trained AI model to generate a draft in real time and sends it to the device. The generated draft reflects the user's writing style and habits.

[1342] Emotion recognition

[1343] The device's built-in emotion engine recognizes the user's emotions by analyzing their typing speed, keystroke patterns, and even facial expressions using a camera.

[1344] Emotion-based writing style adjustment

[1345] The emotion engine adjusts the style and language of the sentences generated by the AI ​​model based on the user's emotional state. For example, if the user is excited, the tone of the sentence will be changed to a calmer tone.

[1346] Real-time feedback

[1347] The device provides real-time feedback reflecting the user's emotions recognized by the emotion engine, allowing the user to see how their current emotional state is affecting their writing and make appropriate corrections.

[1348] View and edit drafts

[1349] The terminal displays the draft received from the server on the user interface. The user can check the displayed draft and make any necessary corrections or additions. The user can edit in real time and the changes are reflected immediately.

[1350] Sending a message

[1351] Once the user has finished editing the message, they click the send button to send the final message. The device then constructs the final message and sends it to the destination via the server in an encrypted connection.

[1352] Specific examples

[1353] For example, consider a case where a user wants to send an important business email under emotional circumstances. The user logs in to the system, collects past business emails, and sends them to the server. The server learns from these and understands the user's writing style and suggested formats. When the user composes a new business email, the system recognizes the user's emotional state and generates a draft like the one below.

[1354] Dear [Recipient Name],

[1355] Thank you for your continued support.

[1356] This is your [username] at [your company name].

[1357] Below we provide additional information regarding the new project we proposed the other day.

[1358] ...

[1359] If you have any questions, please feel free to contact us.

[1360] Sincerely,

[1361] [User Name]

[1362] If the user is in a state of agitation, the system can adjust the tone of the sentence to a calmer and clearer tone, allowing the user to efficiently compose high-quality messages and ultimately achieve effective communication.

[1363] The above is a specific embodiment for carrying out the present invention. This system significantly reduces the effort required for writing text and enables natural and effective communication that takes into account the user's emotional state.

[1364] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1365] Divide the program's processing flow into processing steps

[1366] Step 1: Linking user data

[1367] Step 2: Preprocessing the data

[1368] Step 3: Feature extraction

[1369] Step 4: Training the AI ​​model

[1370] Step 5: Draft generation

[1371] Step 6: Recognize emotions

[1372] Step 7: Adjust your writing style based on emotion

[1373] Step 8: Real-time feedback

[1374] Step 9: View and edit your draft

[1375] Step 10: Sending a message

[1376] Specific explanation of each step

[1377] Step 1: Linking user data

[1378] A user logs in to the system.

[1379] The user enters an email address and password to log in to the system and is authenticated.

[1380] Input: Email address, password

[1381] Output: User's authentication token

[1382] The device is linked to the user's email and chat accounts.

[1383] With the user's permission, the device accesses email and chat accounts through an API.

[1384] Input: Authentication token, account information

[1385] Output: Access rights

[1386] The collected data is sent to a server.

[1387] The device encrypts the collected past emails and chat history and sends it to the server.

[1388] Input: Past communication data

[1389] Output: Encrypted data sent to the server

[1390] Step 2: Preprocessing the data

[1391] The server preprocesses the data received.

[1392] The server cleans the data it receives, removing noise and HTML tags, and then tokenizes it to convert it into a format that is easier to analyze.

[1393] Input: Received communication data

[1394] Output: Preprocessed data

[1395] Step 3: Feature extraction

[1396] The server extracts features from the preprocessed data.

[1397] The server uses natural language processing technology to extract features such as the user's unique vocabulary, punctuation patterns, and line break habits.

[1398] Input: Preprocessed data

[1399] Output: Extracted features

[1400] Step 4: Training the AI ​​model

[1401] The server uses the extracted features to train an AI model.

[1402] The server uses the Transformer model as a machine learning algorithm to evaluate and optimize the model.

[1403] Input: extracted features

[1404] Output: A trained AI model

[1405] Step 5: Draft generation

[1406] The user accesses a new message composition screen.

[1407] The user logs into the system interface using a terminal and opens a message composition screen.

[1408] Input: User interface access

[1409] Output: Message creation screen

[1410] The terminal sends a request to the server to generate a draft.

[1411] The device sends a request to the server to generate a draft using the trained AI model.

[1412] Input: Draft generation request

[1413] Output: A draft creation request is sent to the server

[1414] The server sends the generated draft to the device.

[1415] The server returns the draft generated by the AI ​​model to the device.

[1416] Input: A trained AI model

[1417] Output: Draft document

[1418] Step 6: Recognize emotions

[1419] The emotion engine installed in the device recognizes the user's emotions.

[1420] The device analyzes keystroke speed, typing patterns and facial expressions using a camera to determine the user's emotions.

[1421] Input: key input data, keystroke patterns, facial expression data

[1422] Output: Recognized user emotion data

[1423] Step 7: Adjust your writing style based on emotion

[1424] The emotion engine adjusts writing style and language based on the emotional state it recognizes.

[1425] The server uses AI models to, for example, change the tone of the text to a calmer tone if the user is in an agitated state.

[1426] Input: Recognized emotion data, draft text

[1427] Output: Adjusted sentence

[1428] Step 8: Real-time feedback

[1429] The device provides real-time feedback reflecting the user's emotions recognized by the emotion engine.

[1430] The device displays emotional feedback on the screen, allowing users to see how their current emotional state is affecting their writing.

[1431] Input: Recognized emotion data

[1432] Output: Emotional feedback to the user

[1433] Step 9: View and edit your draft

[1434] The terminal displays the draft received from the server on the user interface.

[1435] The device provides an interface that allows you to edit the displayed draft as it is.

[1436] Input: Draft text

[1437] Output: Editable interface

[1438] The user checks the displayed draft and makes any necessary corrections or additions.

[1439] Users can edit in real time and see changes reflected immediately.

[1440] Input: Draft text

[1441] Output: Final text with corrections and additions

[1442] Step 10: Sending a message

[1443] The user clicks the send button to send the message.

[1444] After the user has completed the modifications, the final message is sent by clicking the send button.

[1445] Input: Final text with corrections and additions

[1446] Output: Send request

[1447] The terminal constructs the final message and sends it to the destination via the server.

[1448] The device encrypts the message according to the destination protocol and sends it to the intended recipient via the server.

[1449] Input: Send request, destination information

[1450] Output: Message sent

[1451] The above are the specific processing steps and their respective operations of this system, which make the writing process more efficient and realize natural communication that reflects the user's emotional state.

[1452] (Application example 2)

[1453] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1454] Customer support plays an important role in modern electronic payment services. However, while prompt and appropriate responses are required for user inquiries and troubleshooting, they often require a great deal of time and effort. Providing an efficient response is particularly difficult when it is necessary to take into account the user's emotional state. The purpose of this invention is to solve these problems and automatically generate prompt and emotionally sensitive responses.

[1455] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for collecting past communication data of a user, means for extracting features from the collected communication data, means for training an AI model based on the extracted features, means for generating drafts of sentences in real time using the trained AI model, means for displaying the generated drafts to the user, means for grasping the user's emotional state, and means for adjusting the style and wording of the generated sentences based on the grasped emotional state. This enables high-quality response sentences to be generated quickly and with consideration for emotions.

[1456] "User's past communication data" refers to the communication history of the user, such as emails and chats that the user has sent and received in the past.

[1457] "Features" are important attributes or patterns extracted from data to analyze the information.

[1458] An "AI model" is a computational model that uses artificial intelligence technology to learn from data and make predictions and classifications.

[1459] "Generating a draft of a text in real time" means creating a first draft of a text based on data that is instantly input.

[1460] "Displaying the generated draft to the user" means displaying the first draft generated by AI on the user's device screen.

[1461] Understanding the "user's emotional state" means observing and analyzing the user's current mental and emotional state.

[1462] "Adjusting style and language" means changing the way you write and express things appropriately.

[1463] An "interface" refers to the operation screen and input device that allow the user to interact with the system.

[1464] "Other people's communication data" refers to communication history such as emails and chats sent and received in the past by individuals other than the user.

[1465] "Imitating a style" means reproducing the writing style or expression of a particular person.

[1466] This invention is a system for automatically generating optimal responses to user inquiries and problems in customer support for electronic payment services. This system takes into consideration the user's emotional state and can generate quick, high-quality responses.

[1467] Overall system overview

[1468] The system primarily uses the following hardware and software components:

[1469] server

[1470] User's device (PC or smartphone)

[1471] Natural language processing libraries (e.g. spaCy or NLTK)

[1472] Transformer models (e.g., BERT and GPT-3)

[1473] Emotion recognition software (e.g., face-api.js)

[1474] Data Collection Module

[1475] When a user logs in to the service, the system collects the user's past communication data, including past emails and chat history. This data is securely transmitted to the server using SSL / TLS encryption protocols.

[1476] Model Learning Module

[1477] The server preprocesses the incoming communication data. This includes cleaning and tokenizing the data using an automated NLP library (e.g., spaCy or NLTK). Next, it extracts features from the data (e.g., wording and punctuation patterns). These features are used to retrain a Transformer model (e.g., BERT or GPT-3) to match the writing style of a specific user.

[1478] Emotion Recognition Module

[1479] The user's device monitors the user's typing speed and patterns in real time, and in some cases analyzes the user's emotional state using facial recognition technology (e.g., face-api.js) using a webcam. This information is used to understand the user's current emotional state (e.g., impatience, anxiety).

[1480] Text generation and style adjustment module

[1481] The trained AI model generates real-time draft responses to user queries, which are then sent from the server to the user's device, while adjusting the writing style and wording based on data obtained from the emotion recognition module.

[1482] User Interface Module

[1483] The generated draft is displayed on the user's screen, and the user can make corrections or additions in real time. The final confirmed message is sent by clicking the "Send" button.

[1484] Examples of specific examples and prompts

[1485] For example, if a user reports a problem with a "payment not being posted," the following response might be generated:

[1486] Dear (Username),

[1487] Thank you for your help. This is (Name of person in charge) from the customer support department of the electronic payment service.

[1488] We will look into the issue of your reported payment not being reflected as soon as possible. We apologize for any inconvenience this may cause.

[1489] Could you please provide the following information:

[1490] Transaction ID at the time of payment

[1491] Payment date and time

[1492] Payment amount

[1493] We will respond promptly based on the information you provide.

[1494] Thank you for your cooperation.

[1495] Sincerely,

[1496] (Name of person in charge)

[1497] If the user is in a state of agitation, the tone of the text will be adjusted to a calmer tone.

[1498] Example prompt sentence:

[1499] "Generate a response for a situation where a user is reporting difficulty with a payment and the user is in a panicked state. Consider the user's past inquiry history."

[1500] This system will significantly improve the efficiency and quality of customer support for electronic payment services.

[1501] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1502] Step 1:

[1503] A user logs in to the service. At this time, the user links their email account and chat account to the system. After linking, the user's device collects past email and chat history from these accounts. The collected data is encrypted and sent to the server. The input is the user's login information and communication data, and the output is the transmission of encrypted data.

[1504] Step 2:

[1505] The server preprocesses the communication data it receives. Specifically, it cleans and tokenizes the data using a natural language processing library (such as spaCy or NLTK). During this process, it removes unnecessary symbols and spaces and splits the data into words. The input is the decrypted communication data, and the output is the preprocessed, clean data.

[1506] Step 3:

[1507] The server extracts features from the preprocessed data. Features include the user's vocabulary, punctuation patterns, and line break habits. A feature extraction algorithm is used for extraction. The input is the preprocessed clean data, and the output is the features.

[1508] Step 4:

[1509] The server uses the extracted features to train an AI model. A Transformer model (e.g., BERT or GPT-3) is retrained with the specific user's data. This process also includes model evaluation and optimization. The input is the features, and the output is the trained AI model.

[1510] Step 5:

[1511] The user's device monitors the user's input speed and patterns in real time to grasp their emotional state. In some cases, a webcam is used to recognize facial expressions and analyze the user's emotions. The input is the user's input behavior and facial image, and the output is data on the user's emotional state.

[1512] Step 6:

[1513] When a user accesses the new message composition screen, the device sends a request to the server to generate a draft. The server uses a trained AI model to generate the draft in real time. The draft is based on the user's past communication data. The input is the user's request, and the output is the initial draft.

[1514] Step 7:

[1515] The server adjusts the writing style and language of the draft generated by the server based on data obtained from the emotion recognition module. For example, if the user is in a state of agitation, the writing style will be made more calm. The input is the initial draft and emotion data, and the output is the adjusted draft.

[1516] Step 8:

[1517] The generated draft is displayed on the user's device. The user can review the displayed draft and make corrections or additions as needed. This correction is done in real time. The input is the adjusted draft, and the output is the final message as modified by the user.

[1518] Step 9:

[1519] After the user completes the modifications, they click the "Send" button to send the final message. The terminal constructs the final message and processes it for sending. The input is the final message modified by the user, and the output is the sent message.

[1520] The specific processing unit 290 transmits the result of the specific processing to the headset type terminal 314. In the headset type terminal 314, the control unit 46A causes the speaker 240 and the display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[1521] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1522] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the headset type terminal 314.

[1523] [Fourth embodiment]

[1524] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1525] 7, a data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[1526] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1527] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control target 443. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the control target 443 are also connected to the bus 52.

[1528] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[1529] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1530] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1531] The control object 443 includes a display device, LEDs in the eyes, and motors for driving the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[1532] Fig. 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Fig. 8, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1533] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1534] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1535] In the robot 414, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1536] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1537] This invention is a system that collects users' past communication data, extracts features, and generates documents using AI technology. This system is designed to improve the efficiency of email and chat document creation and generate consistent, high-quality documents.

[1538] 1. Data Collection Module

[1539] User data integration

[1540] Users log in to the system and link their email and chat accounts. After linking, the user's device collects past email and chat history from these accounts and sends it to the server. Data transmission is encrypted to protect privacy.

[1541] 2. Model Learning Module

[1542] Data Preprocessing

[1543] The server preprocesses the communication data it receives. Specifically, it cleans and tokenizes the text to extract features such as the user's vocabulary, punctuation, and line break habits.

[1544] Training an AI model

[1545] The features are used to train an AI model. The server uses a machine learning algorithm (e.g., a Transformer model) to build a model that generates sentences that match the user's writing style.

[1546] 3. Sentence Generation Module

[1547] Draft Generation

[1548] When a user accesses the new message creation screen, the device sends a request to the server to generate a draft. The server uses a trained AI model to generate a draft in real time and sends it to the device. The generated draft reflects the user's writing style and habits.

[1549] 4. User Interface Module

[1550] View and edit drafts

[1551] The terminal displays the draft received from the server on the user interface. The user can check the displayed draft and make corrections or additions as necessary. The corrected message is completed on the terminal without being sent to the server again.

[1552] Sending a message

[1553] After the user has completed the edits, the final message is sent by clicking the "Send" button. The terminal then performs the process to send the final message.

[1554] Specific examples

[1555] For example, consider a situation where a business user wants to send an email proposing a new project. The user uses the system and uploads hundreds of past proposal emails to the system. The system learns from these emails and understands the user's writing style and proposal format. When the user creates a new proposal email, the system generates a draft like the one below.

[1556] Dear [Recipient Name],

[1557] Thank you for your continued support.

[1558] This is your [username] at [your company name].

[1559] Below we provide additional information regarding the new project we proposed the other day.

[1560] ...

[1561] If you have any questions, please feel free to contact us.

[1562] Sincerely,

[1563] [User Name]

[1564] Users can review this draft, add specific project details and dates, and then send it, allowing them to efficiently create consistent, high-quality messages.

[1565] The above is a specific embodiment for carrying out the present invention. This system significantly reduces the effort required for writing text and enables natural and effective communication.

[1566] The processing flow will be explained below.

[1567] Step 1:

[1568] A user logs in to the system and links their email account or chat account to the system. The user then enters authentication information to allow the linking.

[1569] Step 2:

[1570] The device retrieves past email and chat history from linked accounts. The device collects this data through an API and encrypts it for privacy reasons.

[1571] Step 3:

[1572] The device sends the encrypted data to the server, which receives it and stores it securely in a database. The data remains encrypted while it is stored.

[1573] Step 4:

[1574] The server retrieves the stored emails and chat history from the database and performs pre-processing, which includes cleaning and tokenization, to ensure the data is in a suitable format for analysis.

[1575] Step 5:

[1576] The server extracts features from the preprocessed data, including the user's vocabulary, punctuation patterns, and line-breaking habits.

[1577] Step 6:

[1578] The server uses the extracted features to train an AI model, using a machine learning algorithm (e.g., a Transformer model) to generate a model that matches the user's writing style. This process also includes model evaluation and optimization.

[1579] Step 7:

[1580] The user opens a new message composition screen. The device sends a request to the server to create a draft, including context information (subject and keywords).

[1581] Step 8:

[1582] The server uses a trained AI model to generate drafts in real time, which reflect the user's writing style and habits.

[1583] Step 9:

[1584] The server sends the generated draft to the terminal, which displays the draft on the user interface.

[1585] Step 10:

[1586] The user can check the displayed draft and make any necessary corrections or additions. The user can make these edits in real time using the device interface.

[1587] Step 11:

[1588] The user clicks the send button on the completed message, and the terminal constructs the final message and processes it for sending.

[1589] Step 12:

[1590] The terminal sends the final message to the server, which then sends the received message to the destination via a mail server or chat server.

[1591] Example 1

[1592] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1593] In today's business environment, users are required to maintain fast, consistent, and high-quality communications. However, manually creating documents is time-consuming and labor-intensive, making it difficult, especially when large volumes of communication are required. Privacy protection and data security are also important issues. Therefore, there is a need for a system that can effectively utilize users' past communication data and generate high-quality documents while safely managing the data.

[1594] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1595] In this invention, the server includes means for collecting past communication data of users, means for extracting features from the collected communication data, means for training an AI model based on the extracted features, means for generating drafts of sentences in real time using the trained AI model, means for displaying the generated drafts to the user, and means for protecting data using encryption technology, thereby enabling users to generate high-quality sentences quickly and consistently, while also ensuring privacy protection and data security.

[1596] "User's past communication data" is a collection of digital data related to communications such as emails and chat messages that the user has sent and received in the past.

[1597] "Features" refer to specific attributes and patterns extracted from collected communication data, including, for example, a user's vocabulary, punctuation, and line break habits.

[1598] An "AI model" is an artificial intelligence framework that has been trained using machine learning algorithms to perform a specific task (in this case, sentence generation).

[1599] "Training" refers to the process of having an AI model learn from collected data and acquire the ability to accomplish a given task.

[1600] A "draft" is an initial version of a message that a user is about to create, and is a basic sentence generated by an AI model.

[1601] "Encryption technology" refers to a technical method used to prevent data from being deciphered by third parties, and is used particularly to increase the security of communications.

[1602] A "user interface" is a display screen or operation panel that allows a user to interact with a system.

[1603] This invention is a system that collects users' past communication data, extracts features, and generates documents using AI technology. This system is designed to improve the efficiency of email and chat document creation and generate consistent, high-quality documents.

[1604] Data Collection Module

[1605] The user logs in to the system and links their email and chat accounts. The device collects past email and chat history from the linked accounts, encrypts it, and sends it to the server. AES encryption technology is used for encryption. The server securely stores the received data.

[1606] Model Learning Module

[1607] The server preprocesses the received communication data by cleaning and tokenizing it, and extracting features such as the user's vocabulary, punctuation, and line-breaking habits. To do this, it uses an NLP (Natural Language Processing) library (e.g., SpaCy or NLTK).

[1608] The server then trains an AI model based on the features. It uses a transformer model (e.g., BERT or GPT) and builds the model on a GPU cloud (e.g., NVIDIA GPU cloud). After the training process, the server creates a sentence generation model that matches the user's writing style.

[1609] Sentence Generation Module

[1610] When a user accesses the new message creation screen, the device sends a request to the server to generate a draft. The server uses a trained AI model to generate a draft in real time and sends it to the device. The generated draft reflects the user's writing style and habits.

[1611] User Interface Module

[1612] The terminal displays the draft received from the server on the user interface. The user can check the displayed draft and make corrections or additions as necessary. In this case, the corrected message is completed on the terminal without being resent to the server.

[1613] Finally, when the user clicks the "send" button, the device processes the final message to send it, allowing users to efficiently create consistent, high-quality messages.

[1614] Specific examples

[1615] For example, consider a situation where a business user wants to send an email proposing a new project. The user uses the system and uploads hundreds of past proposal emails to the system. The system learns from these emails and understands the user's writing style and proposal format. When the user creates a new proposal email, the system generates a draft like the one below.

[1616] Dear [Recipient Name],

[1617] Thank you for your continued support.

[1618] This is your [username] at [your company name].

[1619] Below we provide additional information regarding the new project we proposed the other day.

[1620] ...

[1621] If you have any questions, please feel free to contact us.

[1622] Sincerely,

[1623] [User Name]

[1624] Users can review this draft, add specific project details and dates, and then send it, allowing them to efficiently create consistent, high-quality messages.

[1625] The above is a specific embodiment for carrying out the present invention. This system significantly reduces the effort required for writing text and enables natural and effective communication.

[1626] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1627] Step 1:

[1628] User login and account linking

[1629] 1. A user logs into the system.

[1630] Input: User ID and password

[1631] Output: Authentication result (success / failure)

[1632] How it works: The user enters their ID and password on the dedicated login screen and clicks the login button. The server performs authentication, and if login is successful, their personal page is displayed.

[1633] 2. Users link their email and chat accounts to the system.

[1634] Input: Account information (email address, API key, etc.)

[1635] Output: Notification of integration success / failure

[1636] How it works: The user selects "Account Linking" from the settings screen and enters their account information. When they click the link button, the device will link their account and display a notification when the linking is complete.

[1637] Step 2:

[1638] Data Collection and Encryption

[1639] 1. Collect past email and chat history from the account linked to the device.

[1640] Input: Linked account information

[1641] Output: Collected communication data

[1642] How it works: The device uses APIs and other protocols to collect past email and chat history from your account. The collected data is temporarily stored on the device.

[1643] 2. The collected data is encrypted and sent to the server.

[1644] Input: communication data

[1645] Output: Encrypted data

[1646] How it works: Your device encrypts data using AES encryption and sends it to the server, which then decrypts it and stores it securely.

[1647] Step 3:

[1648] Data Preprocessing

[1649] 1. The server preprocesses the data it receives.

[1650] Input: Received encrypted data

[1651] Output: Preprocessed text data

[1652] How it works: The server decrypts the encrypted data and cleans it, specifically removing unnecessary metadata and duplicate messages.

[1653] 2. The server tokenizes the text and extracts features.

[1654] Input: Cleaned text data

[1655] Output: Tokenized data and features

[1656] How it works: The server uses an NLP (natural language processing) library (e.g., SpaCy or NLTK) to tokenize the text and extract features such as the user's vocabulary, punctuation, and line-breaking habits.

[1657] Step 4:

[1658] Training an AI model

[1659] 1. The server trains an AI model based on the features.

[1660] Input: Feature data

[1661] Output: A trained AI model

[1662] How it works: The server uses a Transformer model (e.g., BERT or GPT) to train an AI model on a GPU cloud (e.g., NVIDIA GPU cloud).

[1663] Step 5:

[1664] Real-time draft generation

[1665] 1. The user accesses the new message composition screen.

[1666] Input: User action (clicking New)

[1667] Output: Draft creation request

[1668] Behavior: The user clicks the "Create New" button on the message composition screen. The device sends a JSON-formatted draft creation request to the server.

[1669] 2. The server generates drafts in real time using a trained AI model.

[1670] Input: Draft generation request

[1671] Output: Generated draft

[1672] How it works: The server processes the draft generation request, generates the draft using the trained AI model, and returns the generated text in JSON format to the device.

[1673] Step 6:

[1674] Check and correct via user interface

[1675] 1. The terminal displays the received draft on the user interface.

[1676] Input: Generated draft data

[1677] Output: Draft text displayed

[1678] Operation: The device parses the received JSON data and displays it in the user interface via a text editor.

[1679] 2. The user reviews the draft and makes corrections or additions as necessary.

[1680] Input: Draft text displayed

[1681] Output: Corrected text

[1682] How it works: A user edits a draft in a text editor and then reviews the completed message.

[1683] Step 7:

[1684] Sending the final message

[1685] 1. The user clicks the "Submit" button.

[1686] Input: User action (click submit)

[1687] Output: Start of transmission process

[1688] Behavior: When the user clicks the "Send" button, the device processes the final message to send it.

[1689] 2. The device sends a message.

[1690] Input: Final modified message

[1691] Output: Message sent

[1692] What it does: Your device sends a message to the other person via email or chat account.

[1693] The above are the specific processing steps of the system program, which enables users to create and send messages efficiently, consistently, and with high quality.

[1694] (Application example 1)

[1695] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1696] Customer support in food delivery services requires quick and consistent responses, which places a heavy burden on customer support agents. Furthermore, when different agents respond, inconsistent writing styles and response policies can lead to lower customer satisfaction. To solve this problem, an efficient and consistent message generation system is needed.

[1697] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1698] In this invention, the server includes means for collecting past communication data of a user, means for extracting features from the collected communication data, means for training an AI model based on the extracted features, means for generating message drafts in real time using the trained AI model, means for displaying the generated drafts to the user, and means for automating customer support in a food delivery service, thereby enabling customer support personnel to efficiently generate consistent, high-quality messages and improving customer satisfaction.

[1699] "User" refers to the customer support staff of the food delivery service.

[1700] "Communication data" refers to records of chat and email exchanges between users and customers that have been collected in the past.

[1701] "Features" refer to the user's vocabulary and structural features extracted from communication data.

[1702] "AI model" refers to an artificial intelligence model that is trained based on features and learns the user's writing style and response patterns.

[1703] "Message Draft" refers to a draft response to a customer that is generated in real time by a trained AI model.

[1704] The "means for displaying" refers to a function for displaying the generated draft of the message on the user's terminal.

[1705] "Food delivery service" refers to all services that involve ordering and delivering meals.

[1706] "Customer response automation" refers to the ability for users to use AI models to automatically provide appropriate responses to customer inquiries and complaints.

[1707] The present invention provides a system for improving the efficiency of customer service in food delivery services and generating consistent, high-quality messages. Specific embodiments of this system are described below.

[1708] 1. Data Collection Module

[1709] The server collects users' past communication data. This communication data is a record of chats and emails exchanged between customer support representatives and customers. The data is encrypted and appropriate measures are taken to protect privacy. The specific software used is an API that uses the requests library to retrieve data from the server.

[1710] 2. Model Learning Module

[1711] The server preprocesses the collected communication data and extracts features, such as stylistic and structural characteristics, from the text. This involves text processing such as tokenization and cleaning. An AI model is then trained using specific machine learning algorithms, such as AutoTokenizer and AutoModelForSeq2SeqLM from the transformers library. This training builds a model that matches the user's writing style.

[1712] 3. Sentence Generation Module

[1713] The server uses a trained AI model to generate message drafts in real time. When a user enters a prompt, the AI ​​model generates an appropriate reply message draft and sends the result to the user's device. This allows for fast, high-quality messages.

[1714] 4. User Interface Module

[1715] The terminal displays the generated draft message to the user. The user can check the displayed draft and make corrections as necessary. The corrected message can then be sent to the customer as is. This allows the user to respond consistently without any hassle.

[1716] Specific examples

[1717] For example, consider the following prompt text entered by a customer support representative:

[1718] (prompt statement)

[1719] "Complaint handling: If a customer reports that there was a foreign object in the food they ordered, how would you respond?"

[1720] Based on this prompt, the AI ​​model generates a draft message like this:

[1721] (Generated Draft)

[1722] "Dear customers,

[1723] Thank you for using our food delivery service.

[1724] We deeply apologize for the report that a foreign object was found in the food you received.

[1725] We will investigate the matter immediately and respond promptly, so please wait a moment.

[1726] We will also give you a special discount on your next order.

[1727] Thank you for your continued support.

[1728] Sincerely,

[1729] Customer Support Representative Name

[1730] The system of the present invention enables customer support personnel to provide efficient and consistent responses, thereby improving customer satisfaction.

[1731] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1732] Step 1: User logs into the system

[1733] The user logs in to the system and links to their food delivery service account. At this time, they enter authentication information such as their ID and password. The system uses the entered authentication information to authenticate the user and allow them to log in. This makes it possible to access the user's past communication data.

[1734] Step 2: Collecting communication data

[1735] The server collects the user's past communication data through the API. Specifically, it uses the requests library to obtain the communication data using the API token and user ID. This data includes chat history with customers and email content. It receives the API token and user ID as input and obtains the communication data in JSON format as output.

[1736] Step 3: Preprocessing the data

[1737] The server preprocesses the collected communication data. This preprocessing includes cleaning and tokenizing the communication data, and uses AutoTokenizer from the transformers library. It receives JSON-formatted communication data as input and obtains clean text data as output. This makes it easier to extract noise-free features.

[1738] Step 4: Feature extraction

[1739] The server extracts features from the preprocessed communication data. Features refer to the user's writing style and structural characteristics, such as word frequency and sentence structure. It receives clean text data as input and obtains a feature vector as output. This feature vector is used for subsequent AI model training.

[1740] Step 5: Training the AI ​​model

[1741] The server trains an AI model based on the features. It uses AutoModelForSeq2SeqLM from the transformers library, receives the feature vector as input, and obtains a trained AI model as output. The trained AI model generates sentences that match the user's writing style.

[1742] Step 6: Enter a prompt statement

[1743] When a user creates a new message, they input a prompt text. For example, they can input a specific prompt text such as "Complaint handling: If a customer reports that a foreign object was found in the food they ordered, how would you respond?" The prompt text is received as input and sent to the system.

[1744] Step 7: Draft your message

[1745] The server generates a message draft in real time based on the input prompt. It uses the transformers library to process the prompt with an AI model. It takes the prompt as input and generates a message draft as output. This draft is adapted to the user's writing style.

[1746] Step 8: View and edit your draft

[1747] The terminal displays the generated message draft to the user. The user can review the displayed draft and make corrections as necessary. The corrected message is saved immediately and is not sent to the server again. It receives the message draft as input and displays it to the user as output.

[1748] Step 9: Sending a message

[1749] After the user completes the modification, the final message is sent. The terminal processes the final message and sends a quick reply to the customer. It receives the modified message as input and sends the final message to the customer as output.

[1750] This process allows customer support representatives at food delivery services to efficiently generate consistent, high-quality messages, improving customer satisfaction.

[1751] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1752] This invention is a system that collects a user's past communication data, extracts features, and generates documents using AI technology. Furthermore, this invention incorporates an emotion engine that recognizes the user's emotions and adjusts the style and wording of the generated text based on those emotions. This system improves the efficiency of email and chat text creation and generates consistent, high-quality text.

[1753] 1. Data Collection Module

[1754] User data integration

[1755] Users log in to the system and link their email and chat accounts. After linking, the user's device collects past email and chat history from these accounts and sends it to the server. Data transmission is encrypted to protect privacy.

[1756] 2. Model Learning Module

[1757] Data Preprocessing

[1758] The server pre-processes the received communication data, which includes cleaning and tokenization, to make the data suitable for analysis.

[1759] Feature extraction

[1760] The server extracts features from the preprocessed data, including the user's vocabulary, punctuation patterns, and line-breaking habits.

[1761] Training an AI model

[1762] The server uses the extracted features to train an AI model, using a machine learning algorithm (e.g., a Transformer model) to generate a model that matches the user's writing style. This process also includes model evaluation and optimization.

[1763] 3. Sentence Generation Module

[1764] Draft Generation

[1765] When a user accesses the new message creation screen, the device sends a request to the server to generate a draft. The server uses a trained AI model to generate a draft in real time and sends it to the device. The generated draft reflects the user's writing style and habits.

[1766] 4. Emotion Engine

[1767] Emotion recognition

[1768] The device's built-in emotion engine recognizes the user's emotions through, for example, the user's typing speed, keyboard usage, and facial expression recognition.

[1769] Emotion-based writing style adjustment

[1770] The emotion engine adjusts the style and wording of the sentences generated by the AI ​​model based on the user's emotional state as recognized by the emotion engine. For example, if the user is excited, the language and style will be changed to a calmer one, and if the user is anxious, the language and style will be changed to a more reassuring one.

[1771] Real-time feedback

[1772] The emotion engine provides real-time feedback to users that reflects their emotions, allowing them to check whether their emotional state is reflected in their writing and make appropriate corrections.

[1773] 5. User Interface Module

[1774] View and edit drafts

[1775] The terminal displays the draft received from the server on the user interface. The user can check the displayed draft and make any necessary corrections or additions. The user can edit in real time.

[1776] Sending a message

[1777] After the user has completed the modifications, the final message is sent by clicking the "Send" button. The terminal will then construct the final message and process it for sending.

[1778] Specific examples

[1779] For example, consider a case where a user wants to send an important business email under emotional circumstances. The user logs in to the system, collects past business emails, and sends them to the server. The server learns from these and understands the user's writing style and suggested formats. When the user composes a new business email, the system recognizes the user's emotional state and generates a draft like the one below.

[1780] Dear [Recipient Name],

[1781] Thank you for your continued support.

[1782] This is your [username] at [your company name].

[1783] Below we provide additional information regarding the new project we proposed the other day.

[1784] ...

[1785] If you have any questions, please feel free to contact us.

[1786] Sincerely,

[1787] [User Name]

[1788] If the user is in a state of agitation, the system will adjust the tone of the sentence to a calmer and clearer tone, allowing the user to efficiently compose high-quality messages and ultimately achieve effective communication.

[1789] The above is a specific embodiment for carrying out the present invention. This system significantly reduces the effort required for writing text and enables natural and effective communication that takes into account the user's emotional state.

[1790] The processing flow will be explained below.

[1791] Step 1:

[1792] A user logs in to the system and links their email account or chat account to the system. The user then enters authentication information to allow the linking.

[1793] Step 2:

[1794] The device retrieves past email and chat history from linked accounts. The device collects this data through an API and encrypts it for privacy reasons.

[1795] Step 3:

[1796] The device sends the encrypted data to the server, which receives it and stores it securely in a database. The data remains encrypted while it is stored.

[1797] Step 4:

[1798] The server retrieves the stored emails and chat history from the database and performs pre-processing, which includes cleaning and tokenization, to ensure the data is in a suitable format for analysis.

[1799] Step 5:

[1800] The server extracts features from the preprocessed data, including the user's vocabulary, punctuation patterns, and line-breaking habits.

[1801] Step 6:

[1802] The server uses the extracted features to train an AI model, using a machine learning algorithm (e.g., a Transformer model) to generate a model that matches the user's writing style. This process also includes model evaluation and optimization.

[1803] Step 7:

[1804] The device's built-in emotion engine recognizes the user's emotions through their typing speed, keyboard usage, and facial expression recognition.

[1805] Step 8:

[1806] The user opens a new message composition screen. The device sends a request to the server to create a draft, including context information (subject and keywords).

[1807] Step 9:

[1808] The server uses a trained AI model to generate drafts in real time, which reflect the user's writing style and habits.

[1809] Step 10:

[1810] Based on the user's emotional state as recognized by the emotion engine, the server adjusts the writing style and language of the generated draft. For example, if the user is excited, the language and style will be changed to a calmer one.

[1811] Step 11:

[1812] The server generates and sends the adjusted draft to the terminal, which displays the draft on the user interface.

[1813] Step 12:

[1814] The user can check the displayed draft and make any necessary corrections or additions. The user can edit in real time.

[1815] Step 13:

[1816] The user clicks the send button on the completed message, and the terminal constructs the final message and processes it for sending.

[1817] Step 14:

[1818] The terminal sends the final message to the server, which then sends the received message to the destination via a mail server or chat server.

[1819] Example 2

[1820] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1821] Conventional writing support systems lack the ability to generate sentences that reflect the user's writing style and emotions, requiring the user to spend time manually correcting and adjusting them. Furthermore, sentences generated without considering the user's emotional state may result in a decline in the quality of communication. A system that solves these issues and generates high-quality sentences in real time that reflect the user's emotions while maintaining the user's writing style is needed.

[1822] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes means for collecting past communication data of a user, means for extracting features from the collected communication data, means for training an AI model based on the extracted features, means for generating a draft of a sentence in real time using the trained AI model, means for displaying the generated draft to the user, and means for recognizing the user's emotions and adjusting the style and wording of the generated sentence based on the user's emotional state. This enables real-time generation of high-quality sentences that reflect the user's style and emotions.

[1823] A "user" is an entity that uses this system to create, manage, and send communication data such as emails and chats.

[1824] "Communication data" refers to the content of emails and chats that a user has sent and received in the past.

[1825] "Features" are analytical units extracted from communication data, such as a user's unique language usage, punctuation patterns, and line break habits.

[1826] An "AI model" is an artificial intelligence algorithm trained based on features, which generates and predicts sentences.

[1827] The "emotion engine" is a function that recognizes the user's emotional state and adjusts the style and wording of the generated text according to that emotional state.

[1828] "Style adjustment" refers to changing the tone and expression of a sentence based on the user's emotional state and characteristics.

[1829] An "interface" is a screen or operating tool that allows a user to interact with the system, and provides functions for displaying, correcting, and sending drafts of text.

[1830] "Real-time" means responding immediately to user operations and immediately reflecting any results or changes that occur.

[1831] "Emotion recognition" refers to the technology or process of determining a user's emotional state based on their typing speed, keyboard usage, facial expressions, etc.

[1832] This invention is a system that collects a user's past communication data, extracts features, and generates documents using AI technology. It also includes an emotion engine that recognizes the user's emotions and adjusts the style and wording of the generated text accordingly. This system improves the efficiency of email and chat text creation and generates consistent, high-quality text.

[1833] User data integration

[1834] The user logs in to the system and links their email and chat accounts to the system. The user authenticates by entering their email address and password. Once the linking is complete, the device collects past email and chat history from these accounts via API and sends it to the server via encrypted communication.

[1835] Data Preprocessing

[1836] The server preprocesses the received communication data, cleansing it to remove noise and HTML tags, and then tokenizes it, breaking down sentences into words and converting them into a format suitable for analysis.

[1837] Feature extraction

[1838] The server extracts features from the preprocessed data, using natural language processing techniques to identify features such as the user's unique vocabulary, punctuation patterns, and line break habits.

[1839] Training an AI model

[1840] The server uses the extracted features to train an AI model, then uses a Transformer model as a machine learning algorithm to evaluate and optimize the model. This process generates the optimal model that best suits the user's writing style.

[1841] Draft Generation

[1842] When a user accesses the new message creation screen, the device sends a request to the server to generate a draft. The server uses a trained AI model to generate a draft in real time and sends it to the device. The generated draft reflects the user's writing style and habits.

[1843] Emotion recognition

[1844] The device's built-in emotion engine recognizes the user's emotions by analyzing their typing speed, keystroke patterns, and even facial expressions using a camera.

[1845] Emotion-based writing style adjustment

[1846] The emotion engine adjusts the style and language of the sentences generated by the AI ​​model based on the user's emotional state. For example, if the user is excited, the tone of the sentence will be changed to a calmer tone.

[1847] Real-time feedback

[1848] The device provides real-time feedback reflecting the user's emotions recognized by the emotion engine, allowing the user to see how their current emotional state is affecting their writing and make appropriate corrections.

[1849] View and edit drafts

[1850] The terminal displays the draft received from the server on the user interface. The user can check the displayed draft and make any necessary corrections or additions. The user can edit in real time and the changes are reflected immediately.

[1851] Sending a message

[1852] Once the user has finished editing the message, they click the send button to send the final message. The device then constructs the final message and sends it to the destination via the server in an encrypted connection.

[1853] Specific examples

[1854] For example, consider a case where a user wants to send an important business email under emotional circumstances. The user logs in to the system, collects past business emails, and sends them to the server. The server learns from these and understands the user's writing style and suggested formats. When the user composes a new business email, the system recognizes the user's emotional state and generates a draft like the one below.

[1855] Dear [Recipient Name],

[1856] Thank you for your continued support.

[1857] This is your [username] at [your company name].

[1858] Below we provide additional information regarding the new project we proposed the other day.

[1859] ...

[1860] If you have any questions, please feel free to contact us.

[1861] Sincerely,

[1862] [User Name]

[1863] If the user is in a state of agitation, the system can adjust the tone of the sentence to a calmer and clearer tone, allowing the user to efficiently compose high-quality messages and ultimately achieve effective communication.

[1864] The above is a specific embodiment for carrying out the present invention. This system significantly reduces the effort required for writing text and enables natural and effective communication that takes into account the user's emotional state.

[1865] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1866] Divide the program's processing flow into processing steps

[1867] Step 1: Linking user data

[1868] Step 2: Preprocessing the data

[1869] Step 3: Feature extraction

[1870] Step 4: Training the AI ​​model

[1871] Step 5: Draft generation

[1872] Step 6: Recognize emotions

[1873] Step 7: Adjust your writing style based on emotion

[1874] Step 8: Real-time feedback

[1875] Step 9: View and edit your draft

[1876] Step 10: Sending a message

[1877] Specific explanation of each step

[1878] Step 1: Linking user data

[1879] A user logs in to the system.

[1880] The user enters an email address and password to log in to the system and is authenticated.

[1881] Input: Email address, password

[1882] Output: User's authentication token

[1883] The device is linked to the user's email and chat accounts.

[1884] With the user's permission, the device accesses email and chat accounts through an API.

[1885] Input: Authentication token, account information

[1886] Output: Access rights

[1887] The collected data is sent to a server.

[1888] The device encrypts the collected past emails and chat history and sends it to the server.

[1889] Input: Past communication data

[1890] Output: Encrypted data sent to the server

[1891] Step 2: Preprocessing the data

[1892] The server preprocesses the data received.

[1893] The server cleans the data it receives, removing noise and HTML tags, and then tokenizes it to convert it into a format that is easier to analyze.

[1894] Input: Received communication data

[1895] Output: Preprocessed data

[1896] Step 3: Feature extraction

[1897] The server extracts features from the preprocessed data.

[1898] The server uses natural language processing technology to extract features such as the user's unique vocabulary, punctuation patterns, and line break habits.

[1899] Input: Preprocessed data

[1900] Output: Extracted features

[1901] Step 4: Training the AI ​​model

[1902] The server uses the extracted features to train an AI model.

[1903] The server uses the Transformer model as a machine learning algorithm to evaluate and optimize the model.

[1904] Input: extracted features

[1905] Output: A trained AI model

[1906] Step 5: Draft generation

[1907] The user accesses a new message composition screen.

[1908] The user logs into the system interface using a terminal and opens a message composition screen.

[1909] Input: User interface access

[1910] Output: Message creation screen

[1911] The terminal sends a request to the server to generate a draft.

[1912] The device sends a request to the server to generate a draft using the trained AI model.

[1913] Input: Draft generation request

[1914] Output: A draft creation request is sent to the server

[1915] The server sends the generated draft to the device.

[1916] The server returns the draft generated by the AI ​​model to the device.

[1917] Input: A trained AI model

[1918] Output: Draft document

[1919] Step 6: Recognize emotions

[1920] The emotion engine installed in the device recognizes the user's emotions.

[1921] The device analyzes keystroke speed, typing patterns and facial expressions using a camera to determine the user's emotions.

[1922] Input: key input data, keystroke patterns, facial expression data

[1923] Output: Recognized user emotion data

[1924] Step 7: Adjust your writing style based on emotion

[1925] The emotion engine adjusts writing style and language based on the emotional state it recognizes.

[1926] The server uses AI models to, for example, change the tone of the text to a calmer tone if the user is in an agitated state.

[1927] Input: Recognized emotion data, draft text

[1928] Output: Adjusted sentence

[1929] Step 8: Real-time feedback

[1930] The device provides real-time feedback reflecting the user's emotions recognized by the emotion engine.

[1931] The device displays emotional feedback on the screen, allowing users to see how their current emotional state is affecting their writing.

[1932] Input: Recognized emotion data

[1933] Output: Emotional feedback to the user

[1934] Step 9: View and edit your draft

[1935] The terminal displays the draft received from the server on the user interface.

[1936] The device provides an interface that allows you to edit the displayed draft as it is.

[1937] Input: Draft text

[1938] Output: Editable interface

[1939] The user checks the displayed draft and makes any necessary corrections or additions.

[1940] Users can edit in real time and see changes reflected immediately.

[1941] Input: Draft text

[1942] Output: Final text with corrections and additions

[1943] Step 10: Sending a message

[1944] The user clicks the send button to send the message.

[1945] After the user has completed the modifications, the final message is sent by clicking the send button.

[1946] Input: Final text with corrections and additions

[1947] Output: Send request

[1948] The terminal constructs the final message and sends it to the destination via the server.

[1949] The device encrypts the message according to the destination protocol and sends it to the intended recipient via the server.

[1950] Input: Send request, destination information

[1951] Output: Message sent

[1952] The above are the specific processing steps and their respective operations of this system, which make the writing process more efficient and realize natural communication that reflects the user's emotional state.

[1953] (Application example 2)

[1954] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1955] Customer support plays an important role in modern electronic payment services. However, while prompt and appropriate responses are required for user inquiries and troubleshooting, they often require a great deal of time and effort. Providing an efficient response is particularly difficult when it is necessary to take into account the user's emotional state. The purpose of this invention is to solve these problems and automatically generate prompt and emotionally sensitive responses.

[1956] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for collecting past communication data of a user, means for extracting features from the collected communication data, means for training an AI model based on the extracted features, means for generating drafts of sentences in real time using the trained AI model, means for displaying the generated drafts to the user, means for grasping the user's emotional state, and means for adjusting the style and wording of the generated sentences based on the grasped emotional state. This enables high-quality response sentences to be generated quickly and with consideration for emotions.

[1957] "User's past communication data" refers to the communication history of the user, such as emails and chats that the user has sent and received in the past.

[1958] "Features" are important attributes or patterns extracted from data to analyze the information.

[1959] An "AI model" is a computational model that uses artificial intelligence technology to learn from data and make predictions and classifications.

[1960] "Generating a draft of a text in real time" means creating a first draft of a text based on data that is instantly input.

[1961] "Displaying the generated draft to the user" means displaying the first draft generated by AI on the user's device screen.

[1962] Understanding the "user's emotional state" means observing and analyzing the user's current mental and emotional state.

[1963] "Adjusting style and language" means changing the way you write and express things appropriately.

[1964] An "interface" refers to the operation screen and input device that allow the user to interact with the system.

[1965] "Other people's communication data" refers to communication history such as emails and chats sent and received in the past by individuals other than the user.

[1966] "Imitating a style" means reproducing the writing style or expression of a particular person.

[1967] This invention is a system for automatically generating optimal responses to user inquiries and problems in customer support for electronic payment services. This system takes into consideration the user's emotional state and can generate quick, high-quality responses.

[1968] Overall system overview

[1969] The system primarily uses the following hardware and software components:

[1970] server

[1971] User's device (PC or smartphone)

[1972] Natural language processing libraries (e.g. spaCy or NLTK)

[1973] Transformer models (e.g., BERT and GPT-3)

[1974] Emotion recognition software (e.g., face-api.js)

[1975] Data Collection Module

[1976] When a user logs in to the service, the system collects the user's past communication data, including past emails and chat history. This data is securely transmitted to the server using SSL / TLS encryption protocols.

[1977] Model Learning Module

[1978] The server preprocesses the incoming communication data. This includes cleaning and tokenizing the data using an automated NLP library (e.g., spaCy or NLTK). Next, it extracts features from the data (e.g., wording and punctuation patterns). These features are used to retrain a Transformer model (e.g., BERT or GPT-3) to match the writing style of a specific user.

[1979] Emotion Recognition Module

[1980] The user's device monitors the user's typing speed and patterns in real time, and in some cases analyzes the user's emotional state using facial recognition technology (e.g., face-api.js) using a webcam. This information is used to understand the user's current emotional state (e.g., impatience, anxiety).

[1981] Text generation and style adjustment module

[1982] The trained AI model generates real-time draft responses to user queries, which are then sent from the server to the user's device, while adjusting the writing style and wording based on data obtained from the emotion recognition module.

[1983] User Interface Module

[1984] The generated draft is displayed on the user's screen, and the user can make corrections or additions in real time. The final confirmed message is sent by clicking the "Send" button.

[1985] Examples of specific examples and prompts

[1986] For example, if a user reports a problem with a "payment not being posted," the following response might be generated:

[1987] Dear (Username),

[1988] Thank you for your help. This is (Name of person in charge) from the customer support department of the electronic payment service.

[1989] We will look into the issue of your reported payment not being reflected as soon as possible. We apologize for any inconvenience this may cause.

[1990] Could you please provide the following information:

[1991] Transaction ID at the time of payment

[1992] Payment date and time

[1993] Payment amount

[1994] We will respond promptly based on the information you provide.

[1995] Thank you for your cooperation.

[1996] Sincerely,

[1997] (Name of person in charge)

[1998] If the user is in a state of agitation, the tone of the text will be adjusted to a calmer tone.

[1999] Example prompt sentence:

[2000] "Generate a response for a situation where a user is reporting difficulty with a payment and the user is in a panicked state. Consider the user's past inquiry history."

[2001] This system will significantly improve the efficiency and quality of customer support for electronic payment services.

[2002] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[2003] Step 1:

[2004] A user logs in to the service. At this time, the user links their email account and chat account to the system. After linking, the user's device collects past email and chat history from these accounts. The collected data is encrypted and sent to the server. The input is the user's login information and communication data, and the output is the transmission of encrypted data.

[2005] Step 2:

[2006] The server preprocesses the communication data it receives. Specifically, it cleans and tokenizes the data using a natural language processing library (such as spaCy or NLTK). During this process, it removes unnecessary symbols and spaces and splits the data into words. The input is the decrypted communication data, and the output is the preprocessed, clean data.

[2007] Step 3:

[2008] The server extracts features from the preprocessed data. Features include the user's vocabulary, punctuation patterns, and line break habits. A feature extraction algorithm is used for extraction. The input is the preprocessed clean data, and the output is the features.

[2009] Step 4:

[2010] The server uses the extracted features to train an AI model. A Transformer model (e.g., BERT or GPT-3) is retrained with the specific user's data. This process also includes model evaluation and optimization. The input is the features, and the output is the trained AI model.

[2011] Step 5:

[2012] The user's device monitors the user's input speed and patterns in real time to grasp their emotional state. In some cases, a webcam is used to recognize facial expressions and analyze the user's emotions. The input is the user's input behavior and facial image, and the output is data on the user's emotional state.

[2013] Step 6:

[2014] When a user accesses the new message composition screen, the device sends a request to the server to generate a draft. The server uses a trained AI model to generate the draft in real time. The draft is based on the user's past communication data. The input is the user's request, and the output is the initial draft.

[2015] Step 7:

[2016] The server adjusts the writing style and language of the draft generated by the server based on data obtained from the emotion recognition module. For example, if the user is in a state of agitation, the writing style will be made more calm. The input is the initial draft and emotion data, and the output is the adjusted draft.

[2017] Step 8:

[2018] The generated draft is displayed on the user's device. The user can review the displayed draft and make corrections or additions as needed. This correction is done in real time. The input is the adjusted draft, and the output is the final message as modified by the user.

[2019] Step 9:

[2020] After the user completes the modifications, they click the "Send" button to send the final message. The terminal constructs the final message and processes it for sending. The input is the final message modified by the user, and the output is the sent message.

[2021] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the control target 443 to output the result of the specific processing. The microphone 238 acquires voice indicating a user input regarding the result of the specific processing. The control unit 46A transmits voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[2022] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[2023] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the robot 414.

[2024] The emotion identification model 59 as an emotion engine may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to an emotion map (see FIG. 9), which is a specific mapping. Similarly, the emotion identification model 59 may determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[2025] FIG. 9 is a diagram illustrating an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and actions arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on the lower side. In this way, in the emotion map 400, multiple emotions are mapped based on the structure by which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[2026] These emotions are distributed in the 3 o'clock direction on emotion map 400, and typically fluctuate between relief and anxiety. In the right half of emotion map 400, situational awareness dominates over internal sensations, resulting in a sense of calm.

[2027] The inside of emotion map 400 represents what is going on in the mind, and the outside of emotion map 400 represents behavior, so the further you go outside emotion map 400, the more visible the emotions become (the more they are expressed in behavior).

[2028] Human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. Emotions can also be created for robots, automobiles, and motorcycles, based on various balances, such as posture and remaining battery life. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. An emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on Voice Emotion Recognition and Emotional Brain Physiological Signal Analysis Systems, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map lists emotions belonging to the "reaction" domain, where sensation is dominant. The right half of the emotion map lists emotions belonging to the "situation" domain, where situational awareness is dominant.

[2029] The emotion map defines two emotions that promote learning. One is a negative emotion on the situation side, around the middle of "repentance" or "reflection." In other words, this occurs when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is a positive emotion on the response side, around "desire." In other words, this occurs when the robot experiences positive feelings such as "I want more" or "I want to know more."

[2030] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​indicating each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple pieces of training data that are combinations of user input and emotion values ​​indicating each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions that are located close to each other have similar values, as in the emotion map 900 shown in FIG. 10. FIG. 10 shows an example in which multiple emotions, "relieved," "calm," and "reassuring," have similar emotion values.

[2031] The system according to the present disclosure has been described above mainly with respect to the functions of the data processing device 12, but the system according to the present disclosure is not necessarily implemented on a server. The system according to the present disclosure may be implemented as a general information processing system. The present disclosure may be implemented, for example, as a software program running on a personal computer or an application running on a smartphone, etc. The method according to the present disclosure may be provided to users in the form of SaaS (Software as a Service).

[2032] In the above embodiment, an example was given in which the specific processing is performed by one computer 22, but the technology of the present disclosure is not limited to this, and the specific processing may be distributed and performed by a plurality of computers including the computer 22. For example, the data generation model 58 may be provided in an external device of the data processing device 12, and data may be generated in the external device in accordance with input data.

[2033] In the above embodiment, an example in which the specific processing program 56 is stored in the storage 32 has been described, but the technology of the present disclosure is not limited to this. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-transitory storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-transitory storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes the specific processing in accordance with the specific processing program 56.

[2034] Alternatively, the specific processing program 56 may be stored in a storage device such as a server connected to the data processing device 12 via the network 54, and the specific processing program 56 may be downloaded and installed on the computer 22 in response to a request from the data processing device 12.

[2035] It is not necessary to store all of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store all of the specific processing program 56 in the storage 32; only a portion of the specific processing program 56 may be stored.

[2036] The hardware resource for executing a specific process can be any of the following processors: An example of a processor is a CPU, which is a general-purpose processor that functions as a hardware resource for executing a specific process by executing software, i.e., a program. Another example of a processor is a dedicated electrical circuit, such as an FPGA (Field-Programmable Gate Array), a PLD (Programmable Logic Device), or an ASIC (Application Specific Integrated Circuit), which is a processor with a circuit configuration designed specifically for executing a specific process. Each processor has built-in or connected memory, and each processor uses the memory to execute the specific process.

[2037] The hardware resource that executes the specific processing may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Also, the hardware resource that executes the specific processing may be a single processor.

[2038] As an example of a system configured with a single processor, first, one processor is configured by combining one or more CPUs and software, and this processor functions as a hardware resource that executes a specific process. Second, there is a system that uses a processor that realizes the functions of an entire system including multiple hardware resources that execute a specific process on a single IC chip, as typified by SoC (System-on-a-chip). In this way, a specific process is realized using one or more of the above-mentioned various processors as hardware resources.

[2039] Furthermore, the hardware structure of these various processors can be, more specifically, an electric circuit that combines circuit elements such as semiconductor devices. The specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps may be deleted, new steps may be added, or the processing order may be rearranged, without departing from the spirit of the invention.

[2040] The above-described description and illustrations are a detailed explanation of the parts related to the technology of the present disclosure and are merely an example of the technology of the present disclosure. For example, the above description of the configuration, functions, actions, and effects is an explanation of an example of the configuration, functions, actions, and effects of the parts related to the technology of the present disclosure. Therefore, it goes without saying that unnecessary parts may be deleted, new elements may be added, or replacements may be made to the above-described description and illustrations within the scope of the gist of the technology of the present disclosure. Furthermore, to avoid confusion and facilitate understanding of the parts related to the technology of the present disclosure, the above-described description and illustrations omit explanations of common technical knowledge that do not require particular explanation to enable the implementation of the technology of the present disclosure.

[2041] All publications, patent applications, and technical standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, or technical standard was specifically and individually indicated to be incorporated by reference.

[2042] The following is further disclosed regarding the above embodiment.

[2043] (Claim 1)

[2044] a means for collecting past communication data of a user;

[2045] A means for extracting features from the collected communication data;

[2046] A means to train an AI model based on the extracted features;

[2047] A means to generate drafts of text in real time using a trained AI model;

[2048] means for displaying the generated draft to the user;

[2049] A system including:

[2050] (Claim 2)

[2051] 10. The system of claim 1, further comprising: means for providing an interface through which a user can modify and submit the generated draft.

[2052] (Claim 3)

[2053] 2. The system of claim 1, further comprising means for learning communication data of a person other than the user and imitating that person's writing style.

[2054] "Example 1"

[2055] (Claim 1)

[2056] a means for collecting past communication data of a user;

[2057] A means for extracting features from the collected communication data;

[2058] A means to train an AI model based on the extracted features;

[2059] A means to generate drafts of text in real time using a trained AI model;

[2060] means for displaying the generated draft to the user;

[2061] a means of protecting data using encryption technology;

[2062] A system including:

[2063] (Claim 2)

[2064] 10. The system of claim 1, further comprising: means for providing an interface through which a user can modify and submit the generated draft.

[2065] (Claim 3)

[2066] 2. The system of claim 1, further comprising means for learning communication data of a person other than the user and imitating that person's writing style.

[2067] "Application Example 1"

[2068] (Claim 1)

[2069] means for collecting past communication data of a user;

[2070] A means for extracting features from the collected communication data;

[2071] A means to train an AI model based on the extracted features;

[2072] A means of generating message drafts in real time using a trained AI model; and

[2073] means for displaying the generated draft to the user;

[2074] A means of automating customer service in food delivery services,

[2075] A system including:

[2076] (Claim 2)

[2077] 10. The system of claim 1, further comprising: means for providing an interface through which a user can modify and submit the generated draft.

[2078] (Claim 3)

[2079] 2. The system of claim 1, further comprising means for learning communication data of a person other than the user and imitating that person's writing style.

[2080] "Example 2: Combining Emotion Engines"

[2081] (Claim 1)

[2082] a means for collecting past communication data of a user;

[2083] A means for extracting features from the collected communication data;

[2084] A means to train an AI model based on the extracted features;

[2085] A means to generate drafts of text in real time using a trained AI model;

[2086] means for displaying the generated draft to the user;

[2087] means for recognizing a user's emotions and adjusting the style and wording of generated text based on the user's emotional state;

[2088] A system including:

[2089] (Claim 2)

[2090] 10. The system of claim 1, further comprising: means for providing an interface through which a user can modify and submit the generated draft.

[2091] (Claim 3)

[2092] 10. The system of claim 1, further comprising means for learning communication data of other people and imitating their writing style.

[2093] "Application example 2 when combining emotion engines"

[2094] (Claim 1)

[2095] a means for collecting past communication data of a user;

[2096] A means for extracting features from the collected communication data;

[2097] A means to train an AI model based on the extracted features;

[2098] A means to generate drafts of text in real time using a trained AI model;

[2099] means for displaying the generated draft to the user;

[2100] a means for grasping the emotional state of a user;

[2101] a means for adjusting the style and language of the generated text based on the perceived emotional state;

[2102] A system including:

[2103] (Claim 2)

[2104] 10. The system of claim 1, further comprising: means for providing an interface through which a user can modify and submit the generated draft.

[2105] (Claim 3)

[2106] 2. The system of claim 1, further comprising means for learning communication data of a person other than the user and imitating that person's writing style. [Explanation of symbols]

[2107] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. a means for collecting past communication data of a user; A means for extracting features from the collected communication data; A means to train an AI model based on the extracted features; A means to generate drafts of text in real time using a trained AI model; means for displaying the generated draft to the user; A system including:

2. 10. The system of claim 1, further comprising means for providing an interface through which a user can modify and submit the generated draft.

3. 2. The system according to claim 1, further comprising means for learning communication data of a person other than the user and imitating that person's writing style.

Images